AWS Cloud Security From the Point of View of the Compliance
Storagemag onlinemay2012 final
1. Managing the information that drives the enterprise
Storage Vol. 11 No. 3 May 2012
DAS lives!
budgets still
challenge
storage shops
also:
friendlier face
for cloud storage
Beware the storage
apocalypse
Big adjustments
to handle big data
The effects of
flash in the cloud
are disk
shortages real?
Cloud Archiving
The best storage application for the cloud just might
be archiving; it’s a great way to keep long-term
data on somebody else’s disks or tapes.
3. editorial | rich castagna
Storage
Cloud storage not
so spooky anymore
Lower prices, no vendor lock-in,
endless storage capacity—cloud storage
O
is getting downright friendly these days.
Friendlier face
for cloud storage
Beware
the storage
apocalypse
ne way you can tell a technology is maturing is when the com-
petition among its purveyors starts to heat up. Cloud storage still
Cloud archiving
has an uphill climb to prove that it’s enterprise-worthy, but the
recent, and almost simultaneous, announcements from Amazon,
DAS lives
Google and Microsoft about lowering their prices suggest it’s get-
ting a lot closer to broader acceptance.
Budgets still
challenge Those three behemoths are the “real” storage behind scores of
storage shops
cloud storage services, along with other providers such as AT&T
Synaptic, Nirvanix and Rackspace. They don’t just command our
Big adjustments
to handle attention because of who they are; we notice because they’re
big data
big enough in the cloud storage market to have the power to ef-
fectively control the direction the technology will take in the next
The effects
of flash in
few years.
the cloud By lowering their prices within days of one another, the Big
Three are clearly squaring off to position themselves well within
Are disk
shortages real?
the sights of enterprise data storage managers. It may also be a
sign the cloud storage market has reached a certain level of ma-
turity, with potential users focusing more on price than the tradi-
tional gotchas of cloud storage (security and bandwidth).
That’s not to suggest that security and bandwidth concerns
3 Storage May 2012
4. Storage
have disappeared; rather, enterprise buyers are likely getting a
little more comfortable with how providers are dealing with those
issues and are moving closer to adding cloud to their storage
arsenals. So, naturally, their attention now turns to price.
What three dominant players do pricewise is bound to affect
the entire market. And if that means making cloud storage more
accessible, that’s good news for enterprises.
Price wars aren’t the only sign that cloud storage might be mov-
ing up the food chain. It’s also looking more and more like “regular”
Friendlier face
for cloud storage storage, meaning most of us have gotten past how exotic the con-
cept of sending data into
Beware the cloud is and are now
the storage
focusing on things that
There are still only a
apocalypse
relate to storage wherever handful of major players,
Cloud archiving and whatever it is. which is always a con-
For example, Nasuni, cern in a market, even if
a vendor of network-
DAS lives
they all drop their prices
attached storage (NAS)
Budgets still appliances that serve as
at the same time.
challenge
storage shops gateways to the cloud,
has been touting a capability it rolled out in early 2011. It’s a
Big adjustments cloud-to-cloud migration facility that didn’t get an awful lot of at-
to handle
big data tention then, but it stands to catch a few more eyes now. What
Nasuni is offering is a way to pull your data out of one cloud stor-
The effects age service and park it at another—without recovering it back
of flash in
the cloud to your data center or even shipping it through some intermedi-
ary. On one hand, it “simply” allows the kinds of data migrations
Are disk
that happen in data storage shops all the time when new gear is
shortages real? trucked in to replace aging systems. But put in a cloud context,
it’s significant in that it may erase another bugaboo about cloud
storage: vendor lock-in.
Yet another sign that cloud storage might finally be escap-
ing its adolescence is cloud storage shoppers starting to evalu-
4 Storage May 2012
5. Storage
ate providers as services rather than “things.” The utility concept
is beginning to take hold, and cloud storage users are more in-
terested in service-level agreements (SLAs) than in what kind of
equipment their data ends up on. “How,” “when” and “why” are all
becoming a lot more important than “where.” If you think of cloud
storage as a big disk in the sky, you might have to alter that im-
age. As cloud storage gains popularity, there’s no way providers
will be able to maintain all their customers’ data on spinning disk.
Tape vendors are salivating at this prospect, and probably laughing
Friendlier face
for cloud storage up their sleeves at the irony of 21st century storage turning out
to be based on a medium that’s been repeatedly declared dead.
Beware Case in point: Fujifilm just launched Permavault, its new cloud
the storage
apocalypse archive service targeting its health care and media and entertain-
ment customers, and behind the curtain is a lot of tape fronted by
Cloud archiving Crossroads Systems’ StrongBox gateway device, which leverages
the Linear Tape File System (LTFS).
DAS lives These are all promising signs, although none is reason enough
to completely forfeit a healthy skepticism about cloud storage
Budgets still just yet. There are still only a handful of major players, which is
challenge
storage shops always a concern in a market, even if they all drop their prices at
the same time. And you still have to use caution when you send
Big adjustments your data off-site, pretty much as you do when you hand your
to handle
big data tapes over to the folks in brown shorts driving big brown trucks.
But even that may be a good sign. The more cloud storage looks
The effects like “traditional” storage, the more likely it will get adopted and
of flash in
the cloud (hopefully) even more vendors will start pushing out products that
will help integrate on-premises storage with cloud services. Inte-
Are disk
grating cloud storage with existing systems—as backup, as an ar-
shortages real? chive, as nearline storage—still seems to be the key to convincing
more storage managers to consider cloud storage services. n
Rich Castagna is editorial director of the Storage Media Group.
5 Storage May 2012
6.
7. storage revolutIOn | jon toigo
Storage
Were those crazy Mayans
thinking about storage?
Give or take a few million years, the Mayans
say we’re doomed; but our data storage systems
O
may be living on borrowed time right now.
Friendlier face
for cloud storage
Beware
the storage
apocalypse
ne reading of the stelae discovered in the ancient ruins in and
around the Yucatan Peninsula holds that the world is kaput as of
Cloud archiving
December 21, 2012. So you can understand why I wanted to get
this column published now.
DAS lives
While the consensus of the scientific community regarding
the Mayan Apocalypse is that somebody did their math wrong by
Budgets still
challenge omitting the exponent that properly places the end of everything
storage shops
at a somewhat later date (41 octillion or 4.124105 x 1028 years af-
ter this December), you just never know. Exponents, or “powers of
Big adjustments
to handle
10” as my first math teacher called them, are shorthand expres-
big data sions after all. As such, they’re simplifications intended to limit
the number of integers required to express large numeric values
The effects
of flash in
so we can do math with our fingers or fit big numbers onto the
the cloud screens of our smartphone calculator applications.
While useful, the incorrect use of exponents can lead to error
Are disk and misapprehension. Instead of our sun going supernova in 50
shortages real?
million years, a misplaced exponent could put this extinction-level
event a mere five years away.
Consider the exponents IDC and others have begun to use to
describe data storage capacity growth. An analyst’s chart pre-
sented at a trade show last year showed storage capacity growth
7 Storage May 2012
8. Storage
worldwide topping 21 exabytes in 2011. That’s 21 x 1018 bytes. Inter-
estingly, the analyst said the data created in 2010 totaled some-
thing like 10 exabytes or 10 x 1018, so all that storage capacity was
a good thing. (For the record, I regard “data explosion” estimates
like IDC’s, funded as they are by array makers, to be just as cred-
ible as the apocalyptic in-
terpretations of the Mayan
stone tablets. Fear, uncer-
Given a 7% to 14% failure
tainty and doubt are use- rate in disk per year…
Friendlier face
for cloud storage ful tools for selling stuff, that means somewhere
whether it’s overpriced stor- between 1.4 exabytes
age arrays or Mayan apoca-
Beware
the storage and 2.9 exabytes of data
apocalypse lypse t-shirts.)
Referring to the stor- will be compromised
Cloud archiving age growth chart, this ana- by simple disk failures
lyst went on to argue that in 2012.
DAS lives transactional data had been
declining as a share of to-
Budgets still tal data being stored, while file data was growing. But that was
challenge
storage shops old news; they had been saying that file storage had exceeded
block storage since the mid-aughties. More interesting to me was
Big adjustments their assertion that the capacity allocated to replicating data had
to handle
big data grown to approximately half of the total capacity deployed, sug-
gesting that most companies were using their most expensive
The effects disk to make copies of the stuff they already stored on their most
of flash in
the cloud expensive disk. If true, this statistic makes me sick to my stom-
ach for three reasons.
Are disk
First, given a 7% to 14% failure rate in disk per year, based on
shortages real? the experience of Google and others, that means somewhere be-
tween 1.4 exabytes and 2.9 exabytes of data will be compromised
by simple disk failures in 2012. It’s a scary thought, and one array
makers use to encourage us to purchase spare drives and unused
capacity to replace failing platters.
8 Storage May 2012
9. Storage
Second, given current estimates of data growth in companies de-
ploying server virtualization—from 300% over the next three years
according to IDC, to more than 600% over the same period per Gart-
ner—the total capacity demand for storing production data will end
up between 300 exabytes and 650 exabytes by 2015. If you double
that number to include disk-based replication schemes, you’re look-
ing at a total data storage capacity requirement that exceeds a
zettabyte (1.3 zettabytes or 1.3 x 1021 by Gartner’s estimate). Factor in
the additional capacity we’ll need to purchase to keep up with drive
Friendlier face
for cloud storage failure rates, and you’ll
need to add another 91 ex-
Beware abytes to 182 exabytes of We need to get strategic
the storage
apocalypse replacement disks. with our storage planning
Third, if you consider or else the apocalypse
Cloud archiving the energy requirements we’ll really be confront-
for that much disk, both
to power and cool them,
ing in the next couple
DAS lives
you’re looking at a sig- of years—perhaps as
Budgets still nificantly greater energy soon as December 21
challenge
storage shops demand and cost than we for some firms—will be
confront today. Hard disk
of our own making.
Big adjustments power consumption rang-
to handle
big data es from approximately 3
watts to 10 watts. Calculate how many disk drives are required
The effects to deliver 1.3 zettabytes of capacity, plus another 100 exabytes of
of flash in
the cloud powered spare drives, and we’re looking at some serious power
consumption. Moreover, the heat dissipation requirements for
Are disk
a storage plant in excess of a zettabyte of capacity will be well
shortages real? above the current estimate of about 2 kilowatts per square foot
of data center floor space to somewhere in the neighborhood of
10 kilowatts per square foot. Mix in the energy required to power
the disk and the energy required to dissipate the heat with the in-
creasing cost of utility power (up 23.2% on average in the U.S. over
9 Storage May 2012
10. Storage
the past two years, according to USA Today), and you’ve created a
real witches brew.
All this paints a pretty apocalyptic picture of data storage and
its costs going forward. Unlike the Mayan Apocalypse, however,
our movement along this path is not pre-ordained or inevitable.
Compression and data deduplication (preferably done as a func-
tion of the file system) will have an impact along the way. And
magnetic media manufacturers are working on reducing power
demands and improving energy efficiency at the component level.
Friendlier face
for cloud storage But altering this dismal picture significantly will require a more
holistic or systemic rethinking of our data storage strategies.
Beware We’ll need to get much more particular about what we store
the storage
apocalypse and where we store it. We’ll need to challenge the disk industry’s
mantra about the inefficacy of tape-based storage and bring it
Cloud archiving back online sooner for hosting, archiving and protecting the 40%
to 70% of data that doesn’t need to be stored on spinning disk. And
DAS lives we might just have to eschew any server virtualization software
approach that requires an unwieldy reconfiguration and replica-
Budgets still tion of our storage infrastructure to obtain anything like accept-
challenge
storage shops able I/O performance from applications.
In short, we need to get strategic with our storage planning or
Big adjustments else the apocalypse we’ll really be confronting in the next couple
to handle
big data of years—perhaps as soon as December 21 for some firms—will be
of our own making and not the result of a galactic reset predicted
The effects by some crazy Mayan text. n
of flash in
the cloud
Jon William Toigo is a 30-year IT veteran, CEO and managing principal of Toigo
Partners International, and chairman of the Data Management Institute.
Are disk
shortages real? Correction: In my January column (“IOPS per what?”), I mistakenly asserted that
HP/3PAR’s 450,000 IOPS record on the Storage Performance Council’s SPC Benchmark
was achieved by short-stroking disk. I was informed this wasn’t the case, as the work-
load was spread across 1,900 drives that weren’t short stroking. While the rig does
support short stroking, the technique wasn’t used in this test. I regret the error.
10 Storage May 2012
11. Server Virtualization: Dream for server admins...
ightmare
N for storage pros.
Get your virtual environment under control.
Check out our Top 10 Server Virtualization Tips for storage managers:
www.SearchVirtualStorage.com/Server_Virtualization
12. Cloud storage
archiving
for
Friendlier face
for cloud storage
One of the best applications for cloud storage
Beware
the storage is data archiving. Cloud archiving services can
apocalypse
offer accessibility and data preservation at a fraction
of the cost of building an on-site archive infrastructure.
Cloud archiving
by phil Goodwin
DAS lives
Budgets still
challenge
storage shops
Big adjustments
to handle
big data
The effects
of flash in It wouldn’t seem necessary to start a discussion about archiving
the cloud
by defining the term, but it is. In the early days of computing, ar-
chiving was understood to be the process of moving data on tape
Are disk
shortages real? to a remote facility for long-term storage. Now, however, archiving
has taken on numerous meanings based on context. Archiving can
be the “auto-archive” simplicity of Microsoft Outlook, moving older
data to cheaper storage as well as more traditional long-term off-
line storage. In the context of cloud computing, we’ll define it to
mean relegating data to a third-party location for the purposes of
12 Storage May 2012
13. Storage
lowering costs, improving data protection or both while still main-
taining a reasonable degree of data access.
How long is long?
Regardless of context, implicit in the notion of archive is time—
typically a long time. But “long” is a relative concept. For most
financial data it means seven years, 20 years for pharmaceutical
research, and more than 50 years for some medical records and
Friendlier face nuclear records. In general, retaining data on spinning (or even
for cloud storage
spin-down) disk for 10 years or more is cost-prohibitive even in the
cloud. So, for the purposes of this discussion, we’ll define “long”
Beware
the storage
as between one year and seven years. For data retention exceed-
apocalypse
ing seven years, disk systems will be the media of choice in only
specialized applications. Some examples of those specialized apps
Cloud archiving include geospatial data (i.e., oil and gas exploration images), medi-
cal images and aircraft maintenance logs where the frequency of
DAS lives access is low but the probability of retrieval at some point is high;
therefore, the time and difficulty of recovering 15-year-old tapes is
Budgets still
challenge
likely to be unacceptable.
storage shops
Price vs. performance
Big adjustments
to handle Cloud-based archive opens the possibility of a “just right” balance
big data
between cost and accessibility. Tape has been, and remains, far
and away the lowest cost method of storing data for years. A typi-
The effects
of flash in cal LTO tape holding approximately 1 TB of data costs roughly $35
the cloud
with monthly off-site storage in the range of 25 cents per month.
There’s no way for even the cheapest cloud disk to compete with
Are disk
shortages real? this price. On the downside, the normal retrieval time for a tape
from archive is next-day delivery plus the time needed to mount
and restore it. This means users will wait about a business day
before being able to access the information requested.
Cloud storage, on the other hand, starts at approximately 10
13 Storage May 2012
14. Storage
cents/GB per month and up (depending on volumes). This adds up
when contemplating hundreds of TBs, but it’s still often less than the
cost to procure, deploy and manage arrays in a central data center.
Whereas tape retrieval is measured in business days, data hosted
on cloud storage can be accessed in seconds. For some apps, this
may be the ideal tradeoff between price and performance.
Cloud advantages, disadvantages
Friendlier face Before going all-in on cloud archiving, however, IT needs to weigh
for cloud storage
the virtues of cloud with in-house archiving. Technologically, cloud
providers can’t offer anything that can’t be implemented in-house.
Beware
the storage
So a company may, for example, choose to implement a tiered
apocalypse
storage infrastructure with
tier 3 high-capacity SATA
Cloud archiving disk to achieve a lower av- IT departments really
erage cost per GB stored. shouldn’t be concerned
DAS lives Generally, organizations will with the underlying
lean toward an in-house
Budgets still solution if they can’t risk
technology, provided
challenge
storage shops the loss of connectivity to a contractual service
remote location, have regu- levels are met.
Big adjustments latory requirements that
to handle
big data require strict data security
oversight or have data retrieval requirements where remote laten-
The effects cy would be unacceptable. This is a fairly restrictive list, but there
of flash in
the cloud are still many applications that are candidates for cloud archiving.
IT organizations can quantify the logistical effort to migrate to
Are disk cloud, but shouldn’t overlook a predictable but unforeseen chal-
shortages real?
lenge: a mind shift from a technology-centric perspective to a
service-level management perspective. IT staff used to making
technology choices and deployments often want to delve into the
cloud vendor’s architecture and “suggest” product or technology-
specific implementations. Rarely are such requests warranted, as
14 Storage May 2012
15. Storage
the vendor maintains full responsibility for managing the cloud
infrastructure. IT departments really shouldn’t be concerned with
the underlying technology, provided contractual service levels are
met. With experience, staff attention will gradually shift from low-
level details to higher-level governance.
Service is the critical factor
Service-level management, then, is critical to the initial decision for
Friendlier face cloud archiving as well as ongoing operations. When shopping for a
for cloud storage
cloud archival vendor, consider the following service-level issues:
Beware
Uptime. For most applications, three nines or four nines of avail-
the storage ability are sufficient to meet business requirements. If you need
apocalypse
five nines, you probably have data access requirements that aren’t
conducive to an archive tier. Data hosted in an archive tier is, by
Cloud archiving
definition, non-critical. The uptime requirement largely determines
how much infrastructure the vendor must provision, so it has a big
DAS lives
impact on the hosting cost. Don’t guess; determine the actual hours
Budgets still
challenge
storage shops
Key cloud archiving considerations
Big adjustments
to handle
big data • loud archiving is a tradeoff between accessibility and cost.
C
It may yield the lowest cost while delivering acceptable data
The effects access performance.
of flash in
the cloud
• sing a cloud provider requires the IT organization to shift
U
from managing machines to managing service levels.
Are disk
shortages real? • learly defined service levels are the key to successful
C
cloud archive hosting.
• rganizations should have an exit strategy in case things
O
go wrong.
15 Storage May 2012
16. Storage
when data will be accessed, access patterns and cost of downtime.
These calculations can be compared to the cost of various uptime
guarantees, and easily justified or rejected based on the compari-
son. Vendors often offer hosting-fee rebates or other performance
penalties for missing service-level agreements (SLAs). However, the
caveats are contained in the fine print, so read them.
Accessibility. Accessibility and uptime aren’t necessarily the
same. The storage may be humming, but the subcomponents ren-
Friendlier face der an application unavailable. If you need redundancy or multiple
for cloud storage
redundancy of data links, for example, you’ll have to pay for them
Beware
the storage
apocalypse
Archive vs. backup
Cloud archiving
while many IT shops still consider their old backup tapes to be
DAS lives “archives,” there are specific use cases and access require-
ments that distinguish archives from backup data. Backups
Budgets still
are done to protect data that’s currently in use; if data has
challenge to be restored from a backup, it generally happens shortly
storage shops
after that backup was made. Backup data typically has a
short shelf life.
Big adjustments
to handle Archives are sets of data that will be retained for a long pe-
big data
riod of time for regulatory compliance, corporate governance
or use as intellectual property. Archives are accessed infre-
The effects
of flash in quently, but are searchable so specific data can be recovered
the cloud
relatively quickly and easily.
The Storage Networking Industry Association makes a dis-
Are disk
shortages real? tinction between cloud backups and cloud archiving services:
“Whereas with Cloud Backup the cloud is simply a reposi-
tory of backup data, with Cloud Archive and Preservation, the
Cloud is where the active processes occur that ensure long
term retention, preservation and viability of data.”
16 Storage May 2012
17. Storage
but the alternative may be unacceptable application outages. Make
sure service levels encompass end-to-end data availability.
Performance. Quantify how many IOPS your applications require
and ensure this number is part of the SLA. IOPS can be measured
either as an average or during peak activity. If you demand IOPS
guarantees at peak, then you’ll have to pay for the vendor to provi-
sion them. Some vendors may offer metered billing, but many orga-
nizations don’t like the potential uncertainty of such billing should
Friendlier face demand suddenly spike. Most organizations will absorb a certain
for cloud storage
amount of constrained operation (especially for an archive tier)
in return for cost certainty. In this case, the SLA is for guaranteed
Beware
the storage IOPS, not absolute performance experienced by the end user. If ap-
apocalypse
plication demands exceed contracted IOPS capacity, it’s rightly the
IT organization’s problem; additional IOPS can always be purchased.
Cloud archiving
Data recoverability. As they do for in-house applications, IT or-
DAS lives
ganizations need to specify recovery point objective (RPO) and re-
covery time objective (RTO) requirements for cloud-based archives.
Budgets still
This is related to uptime, but also covers contingencies such as
challenge
storage shops
data corruption or a component failure that doesn’t affect overall
uptime but impacts individual applications. The vendor should have
default values for RPO and RTO, which may be sufficient for an ar-
Big adjustments
to handle chive tier. Again, don’t guess. Know what kind of data loss and ap-
big data
plication unavailability the business units can financially tolerate.
In many cases, it’s much more than is intuitive.
The effects
of flash in
the cloud Disaster recovery (DR). If the cloud archive is used as off-site
replicated storage to satisfy data redundancy requirements, it
Are disk may not be necessary to consider a DR strategy for this tier. But
shortages real?
buyer beware: Most hosted storage doesn’t include any DR contin-
gency. If the hosted data is “live” data provisioned as hybrid cloud
storage, then a DR plan may be necessary. Hosting providers may
regularly back up the data, but they generally don’t rotate the data
off-site, and if they do, they do so infrequently (e.g., monthly). Al-
17 Storage May 2012
18. Storage
though a disaster at a SAS-70 compliant data center is unlikely, it’s
not impossible. DR capability from a hosting company is often a
significant additional expense and can change the economics of
hosting in a hurry. Make sure data isn’t left in a vulnerable state.
Backup and recovery. Even if the hosting vendor backs up the
data regularly and rotates it off-site frequently, IT organizations
may not be out of the woods. Hosting companies usually have a
limited number of backup software options and tape technolo-
Friendlier face gies. This means their backup format (hardware, software or both)
for cloud storage
may be incompatible with your IT systems. If an IT organization is
forced to do a recovery from the vendor’s tapes, there could be a
Beware
the storage substantial delay in acquiring the necessary infrastructure. Ensure
apocalypse
there’s a way out in a worst-case scenario.
Cloud archiving
Compliance. Archived data that requires special compliance
treatment may still be a candidate for cloud hosting. You’ll need to
DAS lives
ensure the data is retained on immutable media, if required. You’ll
probably also need assurance that strict access guidelines are fol-
Budgets still
lowed and auditable; SAS-70 providers should have such processes
challenge
storage shops
in place.
Cost certainty and granularity. One of the key benefits to
Big adjustments cloud storage hosting for archiving rather than using in-house in-
to handle
big data frastructure is that you pay only for the storage consumed. The
metering should go up or down with use, though it may have a
The effects floor minimum.
of flash in
the cloud
Turn tapes into cloud archives
Are disk
shortages real?
It’s clear that cloud archiving may be attractive to companies with
aging data stored on relatively expensive in-house arrays. More
questionable is whether or not converting from tape-based ar-
chives to cloud archives makes sense. Larger organizations may
have tens of thousands of tapes in off-site archives. The process
of retrieving all those tapes and reading them onto a cloud archive
18 Storage May 2012
19. Storage
infrastructure is daunting. It also assumes the provider has the
necessary hardware to read all the tapes, some of which may be in
obsolete formats. Moreover, there’s no way a cloud provider could
host such a data volume at anything close to the cost of tapes sit-
ting in a glorified warehouse. Disk compression and data dedupli-
cation can help significantly, but the difference in cost is still likely
to amount to a substantial premium.
Even though the hurdles for converting tape to cloud archiving
are high, it may still be a consideration. Tapes more than seven
Friendlier face
for cloud storage years old are likely to be very expensive—and possibly problemat-
ic—to restore. Best practices
Beware dictate that organizations It’s clear that cloud
the storage
apocalypse retrieve and rewrite tapes
every five years to ensure
archiving may be
Cloud archiving the data is readable and attractive to companies
the format is current. It’s with aging data stored
DAS lives a task to be reckoned with. on relatively expensive
For example, with a 10,000
in-house arrays.
Budgets still tape archive and a five-year
challenge
storage shops refresh cycle, a company More questionable is
would have to refresh 2,000 whether or not
Big adjustments tapes each year. That comes converting from
to handle
big data to approximately eight tapes tape-based archives
per workday, which is do-
able, but requires a year-
to cloud archives
The effects
of flash in
the cloud around effort for what’s makes sense.
fundamentally a nonproduc-
Are disk
tion exercise. Here again, the crux of the matter lies in the prob-
shortages real? ability of retrieval. Some organizations choose to allow tapes to
become obsolete in the vault with the knowledge that a recovery
would be painful, but the probability of needing to restore the data
is low enough to be worth the risk. On the other hand, if you know
a recovery is all but inevitable, you may opt to incur the time and
19 Storage May 2012
20. Storage
expense of moving from tape to cloud now, thus saving significant
time and effort later, perhaps under urgent conditions.
That’s not to suggest that tape is losing its role in archiving. It’s
still the lowest cost choice for most situations. In addition, LTO’s
Linear Tape File System (LTFS) is enabling tape to take on a new
role as “tier 4” storage, so it can act as another tier in the cloud (or
data center) that’s provi-
sioned along with tiers 0, 1, Regardless of context,
2 and 3. In a cloud archive
Friendlier face implicit in the notion of
for cloud storage environment, this would
effectively enable a hybrid
archive is time—typically
Beware cloud that offers relatively a long time.
the storage
apocalypse fast access (e.g., minutes)
but at the price point of tape for rarely accessed data. The tapes will
Cloud archiving also have built-in compression, and the options of encryption and
WORM. Using automated tiering software, data can be moved auto-
DAS lives matically to the archive tier.
Budgets still The inevitable “what if”
challenge
storage shops So far, we’ve painted a fairly positive picture of cloud archiving ser-
vices. Usually the effort yields the desired result, but not always.
Big adjustments Organizations should consider what would happen if they trans-
to handle
big data ferred tens of TBs of data to a provider and then failed to realize
the desired or contracted results. Sure, penalties might kick in, but
The effects small monetary penalties wouldn’t fully compensate for the true
of flash in
the cloud cost, aggravation or damage to the IT organization’s reputation
for delivery. Contingencies begin with a contract that may be ter-
Are disk minated without penalty for failure to meet specific performance
shortages real?
levels. It should also include a plan for alternative hosting capabili-
ties, either back in-house or with another provider. Cloud archiving
is fairly low on the list of risky endeavors, but smart organizations
will be prepared for anything. n
Phil Goodwin is a storage consultant and freelance writer.
20 Storage May 2012
21. Data Deduplication:
Fad, fixture... or just a nice feature?
Find out the benefits, drawbacks and functions of this technology with our
Top 10 Tips on Data Deduplication:
www.SearchStorage.com/Data_Deduplication
22. DAS
Friendlier face
for cloud storage
Beware
the storage
apocalypse
Cloud archiving
lives
Direct-attached storage may seem passé,
but it’s making a comeback and gaining
widespread interest. By George Crump
DAS lives
Budgets still
challenge
storage shops
Big adjustments
to handle
big data
Direct-attached storage (DAS) is storage installed in a server or
The effects
of flash in external cabinet that’s still connected directly to the server. DAS is
the cloud
storage that’s essentially captive to a particular server, so the serv-
er doesn’t need to traverse a network to read and write data.
Are disk
shortages real? DAS has been criticized as an inefficient way to connect stor-
age to a server and as an obstacle to the data protection process.
Storage that’s locally attached can’t be shared, which can lead to
a situation where one server can be running out of disk capacity
while others have plenty to spare. Without shared storage, there’s
22 Storage May 2012
23. Storage
no way to balance capacity demands.
DAS could complicate the data protection process because each
server would have to be backed up individually and the data copied
across the network. Alternatively, each server would have its own
locally attached tape device and backup application that would
add even more complexity to the backup process.
Shared storage in the form of a storage-area network (SAN) or
network-attached storage (NAS) device was supposed to address
these issues and thus hasten the extinction of DAS. But DAS is still
Friendlier face
for cloud storage a common method of attaching storage to a server; in fact, it’s
enjoyed something of a comeback in recent years. The resurgence
Beware reached new heights this year when EMC announced a PCI Express
the storage
apocalypse (PCIe)-based solid-state storage product designed to enable its
networked storage systems
Cloud archiving to store some data locally
on the server.
DAS is still a common
DAS lives
method of attaching
SAN and NAS underdeliver storage to a server;
Budgets still
challenge
One reason DAS continues to in fact, it’s enjoyed
storage shops live on is that SAN and NAS something of a come-
have largely underdelivered
back in recent years.
Big adjustments on their promises. SANs were
to handle
big data supposed to make it easy to
create a global pool of storage that could be dynamically divvied
The effects up among servers so that only the capacity actually needed at the
of flash in
the cloud time was assigned to a server. For the first eight years or so of the
technology’s existence, this capability was largely unavailable, and
Are disk SAN storage had to be hard partitioned to individual servers. When
shortages real?
a server needed more capacity, a new partition had to be allocated
to that server and then concatenated into the existing storage
pool on the server or, worse, managed separately. The process of
adding storage to a server on a SAN was very similar to the prior
DAS methodology.
23 Storage May 2012
24. Storage
Data protection was also supposed to get a lot easier. The goal
was to back up the SAN directly and not have to back up the indi-
vidual servers. While a few software applications were able to ac-
complish that feat, all suffered from blindly backing up data and
not understanding what that data was. Users quickly realized they
needed something called “application awareness” to back up ac-
tive applications and then perform intelligent restores. As a result,
some form of backup software was required on the servers.
Finally, the price of SAN or NAS technology is still significantly
Friendlier face
for cloud storage higher than DAS. Many users have decided it’s less expensive to
inefficiently directly attach storage than to efficiently share it.
Beware To be fair, modern SAN
the storage
apocalypse and NAS implementa-
tions have addressed the One reason DAS continues
Cloud archiving early storage allocation to live on is that SAN and
shortcomings with tech- NAS have largely underde-
DAS lives nologies like thin pro- livered on their promises.
visioning. However, the
Budgets still time it took to deliver on
challenge
storage shops the allocation promise allowed DAS to build on its foothold in the
data center. But the other challenges remain, for the most part.
Big adjustments The primary driver for SAN/NAS adoption has been the advent
to handle
big data of server and desktop virtualization, since the ability to move vir-
tual server images between physical hosts requires shared stor-
The effects age. Virtualization also makes application-aware, off-host backup
of flash in
the cloud viable due to the entire server being a file that can be backed up
without interacting with the original physical host. But despite this
Are disk
new and important use case for shared storage, DAS continues to
shortages real? live on in the data center. And its value is increasing.
DAS boot
One of the key reasons for DAS’s continued popularity in the data
center is the need for a local boot drive. While most SANs support
24 Storage May 2012
25. Storage
some form of booting methodology, it still requires specialized host
bus adapters (HBAs) and specific support on the SAN storage sys-
tem. As a result, most physical servers still boot from DAS storage.
Thanks to solid-state drives (SSDs), booting from the local server
offers some specific advantages over booting from the SAN. First,
servers can now be booted or re-booted in seconds from a local
SSD. And the SSD can be used as a virtual memory paging area,
which is incredibly important in virtual environments. As hosts in
these environments get loaded up with virtual machines (VMs),
Friendlier face
for cloud storage they can quickly run out of RAM and begin to use local storage as
a memory paging area. If this local storage is hard disk, perfor-
Beware mance can degrade substantially. When this local storage is mem-
the storage
apocalypse ory based, like flash SSD, the drop in performance is negligible. SSD
as a boot drive allows for
Cloud archiving more virtual machines with-
out the need to purchase Solid-state storage also
DAS lives expensive RAM. plays another role in
the resurgence of DAS
Budgets still Extending the SAN with DAS adoption: as an extension
challenge
storage shops Solid-state storage also to the SAN.
plays another role in the re-
Big adjustments surgence of DAS adoption:
to handle
big data as an extension to the SAN. Leveraging even higher performing
PCIe-based solid-state storage, architectures are now develop-
The effects ing that allow the tiering or caching of data directly to the server
of flash in
the cloud needing it. PCIe SSDs can communicate directly with the CPU and
don’t get bogged down by SAS or SATA protocols like typical SSDs.
Are disk This again makes an ideal virtual memory paging area for RAM-
shortages real?
constrained systems, but it’s the tiering or caching use case that’s
becoming increasingly interesting.
With this architecture, storage systems can intelligently pre-
stage the most active data within the PCIe SSD. Then, when a re-
quest for data is made by an application or user, it will be available
25 Storage May 2012
26. Storage
for high-speed delivery on the PCIe SSD. This means the applica-
tion or user doesn’t have to wait for the request to travel across
the storage network, be accepted and processed by the storage
controllers, wait for hard drives to rotate into position and then
send the requested data or write acknowledgment all the way
back up that infrastructure.
If successful, this model of
storage architecture design Vendors like Nutanix
Friendlier face
would turn the SAN world up- offer products that
side down. Storage on the SAN
for cloud storage
are clusters of servers
would become the central re-
Beware pository of information that’s
with internal storage
the storage
apocalypse growing cold and the local to provide a turnkey
PCIe SSD DAS would be used cloud compute-type of
Cloud archiving for the most active data. The infrastructure suitable
SAN would be used for long-
term retention or backup, and
for more traditional
DAS lives
the server would be used for data centers.
Budgets still active processing. This would
challenge
storage shops lead to SAN storage system
designs where capacity is the focus and performance is less im-
Big adjustments portant. But the one downside to native PCIe SSDs is that you can’t
to handle
big data boot from them, so a local SAS hard drive or even an SSD in a drive
form factor would still be required.
The effects
of flash in
the cloud Cloud compute infrastructure
Other key drivers for the revival of DAS are the designs of massive
Are disk storage environments like those of Facebook, Google and others.
shortages real?
These systems combine compute and storage on a single server
that’s highly networked for communication with the other servers.
These systems often have locally attached storage and the ability
to access data on other servers. They can even leverage a com-
bination of PCIe SSD and hard disk drive (HDD) for booting. These
26 Storage May 2012
27. Storage
online providers and Internet technology companies chose this de-
sign so they could get incredibly cost-efficient architectures with
the ability to scale easily as new servers were added.
This model of DAS converged with compute was thought to be a
limited use case, one that only companies with large online apps
would deploy. Now, however, thanks again to server virtualiza-
tion, there’s often a need to build
scalable compute and storage
Friendlier face
infrastructure simultaneously. Using SSD DAS as a
for cloud storage Vendors like Nutanix offer prod- booting and paging
ucts that are clusters of servers
device can complete
Beware with internal storage to provide
the storage
apocalypse a turnkey cloud compute-type of the storage perfor-
infrastructure suitable for more mance picture.
Cloud archiving traditional data centers.
Server virtualization still needs
DAS lives shared storage to move virtual machine images and provide high
availability. These converged architectures automatically copy
Budgets still data to the other nodes in the cluster so that the virtual machines’
challenge
storage shops images are available to any node in the cluster. This “shared DAS”
model provides the simplicity and cost effectiveness of local stor-
Big adjustments age while providing many of the benefits of a SAN.
to handle
big data
If DAS lives, is SAN dead?
The effects DAS isn’t just living, it’s thriving. There are many storage experts
of flash in
the cloud who believe the data center is moving toward a “DAS mostly” en-
vironment, as described above, where the SAN would become the
Are disk long-term repository while truly active data gets stored locally on
shortages real?
the server that needs it. The software to manage this movement of
data is maturing quickly and will be used to keep active data locally.
It will also be able to acknowledge the writing of new data locally
and then sync that data to the capacity SAN in the background.
The drivers for a potential shift to this “DAS mostly” model are
27 Storage May 2012
28. Storage
the performance demands of the virtual environment and the per-
formance capabilities of solid-state storage. One driver has a need
for data locally and the other has the ability to leverage local data
by avoiding the latency caused by the storage network.
Still lots of storage options
As always, there are a lot of potential options for a storage ad-
ministrator when dealing with storage challenges. The first step
Friendlier face is to invest in a performance analysis tool that can help fine-tune
for cloud storage
the current environment. This maximizes the current investment
and allows for an informed decision when selecting what step to
Beware
the storage
take next.
apocalypse
If the network or storage infrastructure can’t be upgraded due
to budget or time constraints, then a valid approach would be a
Cloud archiving strategy of mixing SSD-based DAS with SAN storage. This would
provide the benefit of improved performance by eliminating the
DAS lives storage network bottleneck for maximum SSD benefit.
If a refresh is in the budget, an investment could be made in
Budgets still
challenge
storage network infrastructure and a shared storage system, such
storage shops as an all-flash device to eliminate storage performance concerns
for the foreseeable future. Still, with this approach, using SSD DAS
Big adjustments as a booting and paging device can complete the storage perfor-
to handle
big data mance picture. n
The effects
George Crump is president of Storage Switzerland, an IT analyst firm focused
of flash in on storage and virtualization.
the cloud
Are disk
shortages real?
28 Storage May 2012
29. Recovery
slows for
storage
shops
Friendlier face
for cloud storage
By Rich
Castagna
Beware
the storage
apocalypse
Cloud archiving Storage budgets have been recovering, but progress
might be slowing. Storage managers are looking for
DAS lives tools to get more out of the gear they have or plan to buy.
Budgets still
challenge
storage shops
Big adjustments
to handle Storage managers have been grappling with spiraling disk capac-
big data
ities and new demands on their data storage infrastructures, while
their budgets have dwindled or languished in the doldrums. As we
The effects
of flash in all strive to regain some measure of predictability, IT planning is still
the cloud
often a tumultuous affair, and storage shops aren’t immune to the
ups and downs of an economy slogging its way to recovery.
Are disk
shortages real? The past few Storage magazine/SearchStorage.com Purchasing
Intentions surveys have indicated that recovery is underway, but our
most recent survey suggests we may still experience a bump or two
on the road to recovery. The good news is that storage managers are
as resourceful as ever and more than willing to entertain new tech-
nologies to meet demands even if their purses are pinched a bit.
29 Storage May 2012
30. Storage
Budgets growing slower Key stat
As the economy swooned in 2009, on a year-over-year
basis, storage budgets fell into negative numbers for
the first time. It took a year for that to turn around with
steady but modest upticks to get back to near-respect-
1.7
petabytes
average amount
able numbers. The latest survey, however, is less encour-
of data that com-
aging. Last year, storage budgets averaged 1.9% higher panies are manag-
than the previous year’s; this time the difference is only ing on all forms
of storage media.
0.8%. Still a positive number, but it appears storage man-
agers won’t have a lot more to spend in 2012.
In dollar terms, the average 2012 storage budget is By the numbers
Friendlier face
for cloud storage approximately $2.7 million, off about 10% from 2011. Big
average company
and midsize companies will do more belt tightening than revenue is $1.3
Beware smaller firms: The average enterprise budget is $7.5 mil- billion, slightly
the storage less than last
apocalypse
lion, down almost 10% from last spring, and midsize bud- year’s $1.4 billion.
gets slipped 8% to $2.4 million. With an average storage
budget of $1.4 million, the picture’s much brighter for Broken down by
Cloud archiving company size, year-
small companies that have struggled as larger firms over-year storage
recovered. budgets have made
only small gains:
DAS lives 1.1% increase for big
companies, 2.3% for
midsize and 0.6% for
Budgets still storage Budget change year over year small companies.
challenge
storage shops 5 Budget money will
3.9 be allocated as it
4 has been for sever-
Big adjustments 3.2 900
3.7 al years, with disk
to handle 3 hardware getting
800
big data 2.9
2
1.8 1.9 the biggest chunk
700
(35%) followed by ADD IN 2012
WILL
1 Spring Fall 0.6 0.8 CURRENTLY INSTALL
staff (14%), soft-
600
The effects ware (13%) and main-
09 09
0
% change
of flash in TB 500 fees (12%).
tenance
the cloud 0% -0.4
Spring Fall Spring Fall Spring Fall Spring Fall Spring
400 52
-1 07 07 08 08 10 10 11 11 12 300
Are disk 200 22 27
shortages real? -2
-1.9 100
-3 106
0
Small Midsi
companies compa
#1
#
30 Storage May 2012
24
31. Storage
Disk storage Key stat
still top priority
Disk systems are still the heart and soul of any storage
operation. Our survey companies reported an overall av-
24%
say price is
erage of 247 TB of data stored on their disks, with large the main factor
companies leading with 706 TB; midsize firms reported in choosing a
primary disk
277 TB and small companies had 106 TB. Those are pretty system vendor.
staggering numbers, which will only grow in 2012 as
companies add an average of 41 TB of disk capacity.
By the numbers
Friendlier face
For the first time since we started asking back in 2008,
for cloud storage Fibre Channel (FC) wasn’t the most popular type of stor- Small companies will
age system installed. With 58% reporting that they’re add 22 TB of capacity,
midsize firms will
Beware running network-attached storage (NAS) gear, those file need another 52
the storage storage systems lead the way, with FC (43%) and iSCSI TB (second-highest
apocalypse hike we’ve ever seen)
(42%) trailing. Looking at plans for 2012 disk-related ex-
and big companies
penditures, the biggest chunk (34%) will once again go will add 89 TB.
Cloud archiving for additional hard drives for existing storage systems—
For respondents
a trend we saw starting back in 2007. New NAS systems planning to buy
DAS lives are the next-highest priority at 17%, with the rest fairly storage arrays,
43% will go with
equally split among FC and iSCSI storage-area networks
midrange systems
(SANs), hybrid systems and direct-attached storage (DAS). while 32% will opt
Budgets still
challenge for low-end systems.
storage shops
Twenty-seven per-
Currently installed disk capacity plus cent will buy DAS
capacity to be added in 2012 for new file storage
Big adjustments
in 2012; the rest will
to handle
big data split their spends
900
89 among NAS filers
800 (19%), NAS gateways
The effects1.8
1.9 700
(15%), virtualiza-
WILL ADD IN 2012 tion (11%), clustered
of flash in 706
the0.6
cloud 0.8 600 CURRENTLY INSTALLED systems (9%) and the
ubiquitous “other”
0 TB 500 (19%).
4
Are disk
Spring Fall Spring Fall Spring
400 52
shortages real?
10 10 11 11 12 300
200 22 277
100
106
0
Small Midsize Big
companies companies companies
#2
31 Storage May 2012
32. Storage
Flash is solid storage Key stat
alternative
Given how slowly storage tends to evolve, the adoption of
solid-state storage is something of a phenomenon, going
65%
of non-users say
5
from luxury trinket to a key component of scores of sys- solid-state
tems in just a 3.9 short years. Right now, 24% of respon-
4 few storage is still
3.2
dents are solid-state drive (SSD) users, with another 10%
too expensive.
900
3.7
3
planning to add it this year. Toss in 29% who say they’re
2.9 800
evaluating the technology and that leaves approximately 1.8 1.9 By the numbers 700
2 WILL ADD I
37% on the SSD sidelines. 0.8Two years ago, only CURRENTL
600
Friendlier face 1 Spring Fall 0.6
for cloud storage The most popular spot for SSD is0909 still in storage arrays
0
% change
TB 500
10% of respondents
0% -0.4
(67%), but 40% have deployed it in servers, 34% in end- said their companies
400
Spring Fall Spring
user systems and 20% inFall caching appliances. Fall Spring Fall Spring were using solid-
Spring The average
Beware -1 07 07 08 08 10 10 11 11 12 state storage vs.
300
the storage capacity of installed solid-state storage is 9 TB—maybe 24% today.
apocalypse
not terribly impressive by hard disk standards, but a hefty
-2
200 22
-1.9 The average1009 TB
of
amount of flash that has grown from an average of 6.8
-3 of installed SSD 106
Cloud archiving 0
TB two years ago. On average, companies planning to add capacity represents Small
solid-state storage in 2012 will up their installed capaci- an increase of 18% companies
since last fall and
DAS lives ties by 7.1 TB. #1 32% vs. one year ago.
Thirty-one percent
Budgets still say they have
challenge 10 TB or more of
Currently using, planning to deploy
storage shops or evaluating solid-state storage SSD installed.
Big adjustments
Using SSD now 24
to handle
big data
10 SPRING 2012
Implementing SSD this year
10 SPRING 2010
6
The effects
of flash in 29
the cloud Evaluating SSD
35
No SSD plans
37
Are disk 49
shortages real?
0% 10 20 30 40 50
15
#3
12.1
10.8 10.8
10
6.4
32 Storage May 2012
5 4.4
33. Storage
3.2 Cloud storage 900
89
Key stat
slowly gaining fans
800
26%
1.8 1.9 700 WILL ADD IN 2012
706
Not too 0.8 ago, it 600 a leap of faith for a company to
CURRENTLY INSTALLED
Spring Fall 0.6 long took
09 09
0 use a cloud storage500
TB service, especially for housing prima-
-0.4 have already
Fall
400 52
Spring Fall Spring or nearline data. But service providers have made great
ry Fall Spring deployed—or
08 10 10 11 11 12
strides in the last couple of years, and they’ve begun to
300 plan to deploy—
277
win over some converts. Today, about 30% of companies
200 22
private storage
cloud products.
-1.9 use some form of non-backup cloud storage (more on
100
106
cloud backup later); the current survey indicates that
0
Small Midsize Big By the numbers
Friendlier face
the top three applications are data center primary data
companies companies companies
#1
for cloud storage (12.1%), disaster recovery (10.8%) and data center nearline Some cloud use is
data (10.8%). To put those numbers in perspective, two substantial: 44%
#2 say they have more
Beware years ago only 14% were using non-backup cloud storage than 10 TB in the
the storage and the top app was DR at 6%. cloud.
apocalypse
Companies using those cloud storage services have
Only 6% are cur-
an average of 19 TB of data stored in the cloud. That’s rently using hybrid
24
Cloud archiving nearly 20% higher than the number reported last fall storage devices,
10 SPRING 2012 when we asked the question for the first time. Current
which integrate on-
premises storage
10 DAS lives SPRING 2010 users also seem pretty pleased with the services as 83% with a cloud
6 plan to add services in 2012, with DR (38%) and storage storage service.
29 for primary data (37%) topping their shopping lists.
Budgets still 35 Nine percent of re-
challenge porting companies
storage shops use a cloud file
37
sharing/synching
49 service; 7% plan to
Does your company use cloud storage services for
10 20 30 40 50 these primary or nearline storage applications? add one this year.
Big adjustments
to handle
big data 15
#3
12.1
The effects 10.8 10.8 SPRING 2012
of flash in
SPRING 2010
the cloud 10
6.4 6.8 6.6
Are disk
shortages real?
5 4.2 4.4
3.3
0%
Data center Disaster Data center Archiving Remote office
primary data recovery nearline data online data
#4
33 Storage May 2012