Your SlideShare is downloading. ×
Informationweekreports doc 2008_210500063
Informationweekreports doc 2008_210500063
Informationweekreports doc 2008_210500063
Informationweekreports doc 2008_210500063
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×
Saving this for later? Get the SlideShare app to save on your phone or tablet. Read anywhere, anytime – even offline.
Text the download link to your phone
Standard text messaging rates apply

Informationweekreports doc 2008_210500063

302

Published on

Published in: Technology, Business
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
302
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
2
Comments
0
Likes
0
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
No notes for slide

Transcript

  • 1. If you have server virtualization in place, you have a powerful disaster-recovery tool at your fingertips LTHOUGH THE ABILITY to leverage virtu- front and center. We’re seeing specialized offerings alized servers for disaster recovery has from established players, like VMware’s Site Recovery always been a core part of enterprise vir- Manager. Citrix XenServer Enterprise and Platinum tualization platforms, until recently it was include Live Migration, and our tests of Microsoft Hy- relegated to the, “oh, by the way” section per-V show that, given sufficient memory, cutting a vir- of most vendors’ feature lists in favor of tual machine over to new hardware is relatively pain- sexier benefits such as reducing server less. Other vendors have launched products or retooledsprawl, cutting capital costs, and, of course, the ever- existing applications with an eye to catching the virtu-present green effect. To be clear, we’re talking disaster alized DR wave. Novell’s PlateSpin Forge and Double-with a capital D—fire, flood, or pestilence, not opera- Take Software’s eponymous offering are examples.tional recovery from user error or file deletion. Most of What wave, you say? A review of our 50 most recentus have daily data protection plans in place. Few have virtualization projects found that nearly 90% of clientsthe mandates, funds, and spunky SAS 70 auditors to that have virtualized their main systems have rolledmake sure our disaster-recovery (DR) strategies in- some level of the technology into their DR plans.volve more than leftover hardware, sketchy testing, and One interesting trend: folks who virtualize their dis-a stale run book. aster-recovery environments but not their primary Now, better DR through virtualization has moved networks. Virtualization enthusiasts may scoff, but this This article, the first in a four-part series, is just one element of a special multimedia package on business innovation. For links to related stories and additional editorial content, go to businessinnovation.cmp.com. Copyright 2008 United Business Media LLC. Important Note: This PDF is provided solely as a reader service. It is not intended for reproduction or public distribution. For article reprints, e-prints and permissions please contact: PARS International Corp., 102 West 38th Street, Sixth Floor, New York, NY 10018; (212) 221-9595 www.magreprints.com/quickquote.asp
  • 2. Special Report BUSINESS INNOVATIONmakes sense on a few levels. It’s much less expensive back to the CEO and COO for clarification on antici-than building a standard DR site. It’s a great way to in- pated usage. That will be a major factor in decisionstroduce virtualization and build up internal skills on bandwidth, host servers, and storage configuration.without affecting the production network. And it skirts Any disaster-recovery plan needs to define mission-the pesky issue of vendors that don’t officially support critical access parameters. A manufacturing or con-virtualized versions of their applications. struction company that relies on IT systems for back- A midsize private school in Rhode Island took this office support may opt to build the DR site to supportroute. After a disaster befell the institution, IT opted to half the normal usage and workload, giving priority tofill a hole in its DR plan using a virtualization appliance department heads or critical apps. If you run cus-rather than virtualizing the production network. The tomer-facing, mission-critical data on your systems,creator of an accounting application critical to the school you’re going to want matching configuration and as-doesn’t (yet) support running the app in a virtualized sociated bandwidth.environment. We proved that the software would run in What happens if you undersize your DR site? A majora VM and set it up at the DR site.When the vendor even- benefit of virtualization and SAN usage is the ability totually adds VM support, the school is a step ahead. quickly expand the supporting hardware infrastructure. For organizations that already have virtualized their Let’s say you have a 10-node host server cluster at theservers, the challenge is deciding on the size of the DR home office. Your DR site holds five servers that wouldsite configuration and setting a failover level. If you’ve support the main office in the event of a disaster. Yourjust implemented virtualization, you’ve likely got a bit plan can include a provision that if a failover lasts forof a budget windfall (which will disappear quickly more than 48 hours, you would add host servers to im-once the CFO figures out what’s going on) and some prove performance and spread the load.unused but perfectly functional gear. So if your pro-duction environment is 275 virtual servers running on THE IMPOSSIBLE BECOMES POSSIBLE30 physical hosts with 10 TB of data on a SAN, do you It doesn’t matter if your platform is from Citrix, Mi-need the same capacity for your DR site? Or can you crosoft, Sun Microsystems,VMware, or Joe’s open sourcewhittle down to fewer servers to save money? emporium—if you have server virtualization, you’ve got How do you decide? You don’t. This is one place IT the basic elements of a cost-effective disaster-recoverymust get business leaders involved. Push the DR plan plan. Separating physical hardware from logical servers Impact Assessment: Virtualized Disaster Recovery Benefit Risk IT Virtualization as a core part of your disaster- Don’t get complacent. Ensure that you tightly organization recovery strategy opens up new data protection manage VM migration and creation, don’t possibilities while maximizing your investment neglect DR drills, and don’t skimp on bandwidth in the technology. between sites. Business Virtualization helps IT create a more robust and Minimal risk, especially if you’re already on the organization flexible DR plan, at a lower cost than traditional virtualization path. The main threat is depending alternatives, minimizing fears of unplanned on a DR site that’s underpowered in terms of outages. server horsepower and bandwidth. Business By fully leveraging the investment in virtualization If IT misses a core legacy or nonvirtualizable competitiveness and designating existing remote offices as DR system, the business could incur a significant sites, the enterprise frees up resources for productivity hit. There’s no substitute for a other, more bottom-line-focused, initiatives. complete inventory of assets. Bottom Line Virtualization is a great way to implement comprehensive disaster recovery at a significantly reduced cost. However, as with any DR strategy, if you don’t commit the resources to plan and test properly, and for staff training, you could face a nasty surprise during a real disaster.35 Sept. 8, 2008 informationweek.com
  • 3. Special Report BUSINESS INNOVATIONand applications means those ultimate objects of CIO tapes and shipping them off-site is simple and is andesire, namely tight recovery point objectives (RPOs) easy check box on the “in case the building blows up”and recovery time objectives (RTOs), are finally attain- checklist. Don’t stop there, though—if there were a dis-able, and without gutting the budget. Digging in and aster, you’d have to rebuild your infrastructure, restoredefining RPOs and RTOs for every major piece of data data, and get everything back online. So rather thanwithin your organization will set the stage for your over- saving just data, backup tapes should now include rel-all DR plan as well as daily backup, replication, and evant virtual machines. Typically, we see clients takearchiving operations. Get consensus on what objectives copies of their base virtualized servers, set them up atmake sense based on estimated costs of lost time and the DR site, then power down. Changes to databasesdata. Not only will this exercise sharpen the focus on DR and data volumes are either replicated or transferred toplanning, it helps justify the expenditure. the site. Failover would involve booting the virtual Next, you need a data replication/backup plan that servers and restoring relevant data from tape or disk.integrates with your virtualized environment. Virtual- A midsize distributor in the Northeast made goodization is about servers and applications, not specific use of the spare equipment that came as a result of itsdata. If you’ve done server virtualization but don’t have virtualization project by building cold standby servers.a SAN, you’re missing out on some of the built-in repli- Rather than simply changing tapes, the company setcation and snapshot features that will help flesh out up a smaller version of its entire network, using serversyour strategy. Finally, make sure your current backup or freed up by virtualizing. The gear was configured,continuous data protection software supports virtual- loaded with the company’s core VMs, then shut downization backup, data replication, and remote restores. and shipped off-site. Data tapes are still rotated on aMajor vendors, including CA, Symantec, and Tivoli, regular basis, but the company has added a “refresh”have expanded or retooled their product lines to pro- plan for updating server images on a biannual basis. Itvide better functionality for virtual data protection. already had the base servers, so costs were limited to additional memory, configuration effort, and licensing.JUST DO IT Many companies are looking at remote offices as Typically, virtualized disaster-recovery environ- “found” DR sites that make the most of existing infra-ments fall into four categories: cold standby, remote structure and telco costs. For example, a New York in-site, hot site, and nonvirtualized office. vestment firm we worked with retooled its DR plan Cold standby is the most popular. Saving raw data to based on its virtualization and storage project, which No Virtualizing Without A License AFETY ISN’T FREE. rection you have failover. For example, if disaster recovery. Servers must be turned Don’t assume you’re you’re configured to fail over in a single off and used only in the event of a disaster. covered with your li- direction only, you need Site Recovery li- This applies to all Microsoft products, but censing when you add a censes for the primary site. If configured it requires that you carry Software Assur- failover host or server. In fact, for bidirectional failover, you need to li- ance on the OS, application, and thoselicensing rules vary widely by vendor, so we cense hosts at both sites. pesky CALs, which can kill a budget sur-asked a specialist, Rob O’Shaughnessy, Citrix Xen has the same rules as VM- plus faster than a bipartisan vote.manager of software licensing at Green- ware around licensing hosts, hot or cold. As you move up the application stack,Pages, for the lowdown on what an organ- One difference: Citrix licenses per server, things get trickier. For example, you canization needs for a disaster-recovery site. regardless of CPUs, whereas VMware li- have an Oracle Enterprise server license VMware specifies that if you’re using censes are per CPU. running on a failover server, but only if ita primary and recovery site, you’ll need With Microsoft, not surprisingly, the pic- accesses the same databases as theto be licensed for Virtual Center Manager ture gets a bit fuzzy. If you don’t have Mi- production server and is never on forServer at both locations. You’ll also need crosoft Software Assurance, you need li- more than 10 days—we kid you not.VMware Infrastructure licenses for any censes for every server you have running, As part of a DR plan, we highly recom-server hosting VMware ESX—whether even cold servers. However, if your organ- mend compiling a full software list and re-it’s running or powered down. ization has purchased Software Assur- viewing it with licensing vendors to ensure If you’re also using the Site Recovery ance, one benefit is complimentary cold that you’re compliant and not paying forManager, you’ll need licenses for any di- backup server licenses for purposes of licenses you don’t need. —MIKE HEALEY Sept. 8, 2008 38
  • 4. employed CA’s XO Soft DIG DEEPER VMware’s virtualizationreplication app in the VM- engine and leverages VIRTUAL STANDSTILL Don’t drive too far into serverware environment for serv- virtualization without considering its impact on your storage PlateSpin’s conversioners and replicated core data infrastructure. Download this InformationWeek Report at: and recovery software.using an EqualLogic SAN informationweek.com/1191/report_san.htm Replication is configuredarray. The result? based on bandwidth and See all our Reports at informationweekreports.com “We were able to create desired recovery win-a complete failover configuration without having to dows. The boxes come in two sizes with the ability tomove toward a hosted site,” says the firm’s IT director. support 10 or 25 servers. The $50,000 price for the 25-“This included replicating all core data and building server edition seems steep, but not when you considerredundancy into our Domino mail system.” that it comes with 2.5 TB of storage and 16 GB of RAM, One concern if you’re pursuing this strategy is includes all virtualization licensing, and supports vir-bandwidth. The financial firm has a 100-Mbps link tualized and nonvirtualized boxes. Nonvirtualizedbetween offices that are open only during normal servers must be configured with PlateSpin’s converterbusiness hours. This gives IT plenty of time for repli- and will require some tweaking of images and settings.cation and updates. If you’ve got a small pipe between A few caveats: At press time, the devices supportedsites, don’t expect to replicate gigabytes of data Windows servers only. However, given PlateSpin’s re-overnight. In theory, a clean T1 can push 550 MB to cent acquisition by Novell, we expect Linux support in650 MB per hour, but that’s best case and leaves scant short order, maybe even a physical-to-virtual utilityroom for error when moving differential data of 40 for NetWare servers. In addition, your investmentGB or more a night. doesn’t buy bandwidth enhancements or speed im- Hot sites generally are maintained only by publicly provements versus replication between servers, andtraded companies, especially financial firms, and those finally, PlateSpin doesn’t address all SAN storage orwho are truly paranoid and flush with cash. Determin- database replication. Zenith Infotech offers a similaring capacity of a hot DR site ties back to RPO and RTO appliance, built on Sun virtualization technology, withmandates. If your recovery point is zero data loss and similar OS and SAN support limitations.recovery time is less than 10 minutes, you’re looking ata major investment at all levels, including bandwidth MAKE YOUR CASE(fiber is a must) and replication (specialized failover, Money talks, so we ran some numbers comparingplus data replication, plus database and mail-specific the cost to set up virtualized and standard DR sites forspecialty apps). At this level, we’re seeing a strong in- a network with 30 servers and 4 TB of storage capacity,terest in VMware’s Site Recovery Manager. Introduced assuming the main production network isn’t virtual-late last year, SRM essentially scripts the entire pro- ized. The lion’s share of savings comes from servercess you’d go through when failing over a site. It’s not a hardware. In a standard setup, mirrored configurationreplication system—you’d still need to use your SAN’s of the main facility, requiring like-for-like hardware,replication utility or a third-party product like CA’s XO would run $150,000. In a virtual scenario, we’d useSoft or Double-Take. But SRM takes over whenever larger servers with additional memory and processors.there’s a major outage, essentially managing the cut- Target consolidation is 10-to-1 at a cost of $30,000, forover of your servers between locations. The system a net savings of $120,000. We tallied software licensingeven has a neat feature to print out a DR run book out- costs at $100,000 for the standard site and $120,000 forlining your plan—auditors love that stuff. the virtualized site, to account for core licenses plus There are a few gotchas: virtualized VMware servers virtualization and associated backup software. Storageonly, of course, plus there’s currently no support for capacity is a wash. We assumed standard site engi-replication of raw data LUNs on a SAN. neering at $40,000, then doubled that for the virtual- ized site to account for the additional work requiredSTART SMALL to perform physical-to-virtual conversions and create For organizations that aren’t virtualized yet but want an update plan. Even without the long-term utility andto use virtualization technology for DR, several ven- other savings represented by maintaining fewer phys-dors offer appliances that leverage virtualization and ical servers, in our example scenario we find the vir-inexpensive storage to build out a cost-effective DR tual setup costs $60,000 less than the standard site.system. One example is PlateSpin’s Forge product,which has grown in popularity as an appliance alter- Mike Healey is CTO of GreenPages Technology Solutions,native to building a mirrored (or smaller version) of www.greenpages.com, and an InformationWeek contributor.your production environment. The system is built on Reach him at michael.healey@greenpages.com.39 Sept. 8, 2008 informationweek.com

×