This document discusses object storage, specifically OpenStack Swift, within enterprise environments, highlighting its advantages over traditional storage types due to data growth and the need for scalability. It explains the architecture of Swift and its components, including how it handles metadata and object retrieval through HTTP APIs. Additionally, it outlines use cases, performance considerations, and the benefits of combining OpenStack Swift with private, public, or hybrid cloud infrastructure.

1. OpenStack Swift In the Enterprise
An Introduction to Object Storage and Swift
1

2. • This webinar is being recorded and an on-
demand version will be available at the
same URL within 15 minutes after we
conclude
• Please submit questions via the button to
the upper left of the viewer
• On Twitter? Join the conversation:
#objectstorage, #swift, #openstack
@HOSTINGdotcom
2
Housekeeping

3. • Critical factors driving new storage types
• What is Object Storage?
• OpenStack Swift Architecture
• Key Considerations
3
Today’s Agenda

4. 4
Critical factors driving new storage
types
• Object storage began when
applications started to store
and serve data to any device
regardless of where they are
connecting from
• Explosion in non-transactional
data (i.e. Machine generated,
rich media content, etc.)
• Cloud architectures
demanding alternatives to
storing and retrieving content

5. Critical factors driving new storage
types cont..
• Retention of data has
increased
• Performance
degrades as number
of files or storage
increases
• IT budgets are not
increasing
5
Data growth is exploding

6. A look at Traditional Storage Types
Block Based File Based
• Sits on top of Block Based Storage
• Stores FilesFolders that are
humanly readable
• Can increase and decrease size of
storage
• NIFSCIFS Protocols allow for a
single namespace
• Scaling limitations: Metadata is
held within the file, performance
suffers as number of files increase
• Is designed for LAN based access,
not Internet based devices
6
• Data is structured with a maximum
length (i.e. block size)
• Typically stored in SAN
environments where data is stored
in volumes
• Can increase storage, however, it
cannot decrease* (*without some
very nifty & complex tooling)
• Scaling the size of a block disk is
subject to vendor and partition
limitations
• Is designed for to be accessed by
low latency, high bandwidth
networking

7. 7
Introduction to Object Storage
• Object storage, often called Cloud
Storage, is commonly used by cloud
providers due to it strength in
dynamically increasing or decreasing
storage demand dictates
• All objects (i.e. movies, pictures,
documents, etc.) are stored with
• A unique Identifier
• Meta Data
• Attributes
• Data
• Is accessed via HTTP restful API
allowing applications to access the data
from anywhere
• The most commonly used Object
Platforms are AWS’s S3, and

8. Object Storage: What
this Enables
• Consider an application where
patient records are inputted
• The application has form fields
where the care providers inputs
diagnosis, patient local,
symptoms, etc. (i.e. Metadata of
a patient record)
• Now imagine the care provider
or research facility wants to look
at this data from not just one
hospital or location but abroad
to learn about trends
• How can this data be accessed
independent of any one facility
while ensuring the security?
Source: RightScale 2014 State of the Cloud Report
• Object Storage allows application to retrieve
metadata of a file from billions objects
• The retrieval of a given object is
authenticated and authorized
• The data can be retrieved over the Internet
or over private networks instantly via
HTTP(s)

9. Swift based Object Storage
9
Feature Benefit
Swift is a core OpenStack Project OpenStack is the most successful open
source project ever that set out to solve
common challenges with cloud
technologies.
Deployment Flexibility Can be deployedconsumed from Public,
Private, Hybrid cloud designs
Unlimited Storage Huge & flat namespace, highly scalable
readwrite access. Ability to serve content
directly to from storage systems
Multi-dimensional Scalability (scale out
architecture)
Backup and archive large amounts of data
with linear performance
Easily* add capacity unlike RAID resize Elastic data scaling with ease
No central database Higher performance, no bottlenecks
Real-time visibility into client requests Know what exactly users are requesting
Numerous ways to access storage Users can access storage via CLI, HTTP,
file-system gateways

10. The Core Components of Swift
10
Swift Proxy
Storage Node
Identity Service
Understands where data lives &
routes requests
Runs account, container, &
object services
Authenticates and authorizes
storage requests

11. Swift Data Model
Three components to
unstructured data
• Account: A unique name
storage area that will
contain the metadata
• Container: User-defined
storage area containing
metadata of the
container and list of
objects
• Object: Is where the
data object and its
metadata is held
Account: Customer A
Container 1
Object
Orange
Object
Green

12. Swift Zones and
Regions
• Zones:
• Replicates data among nodes
within the environment
• Default replica is 3
• Ensure no networking or
hardware single point of failure
• Regions:
• Conceptually designed to
distribute replicas among
geographically dispersed
areas
• Used for offsite DR purposes
• Used in ActiveActive Multi-
sharing environments
Region A Region B
Zone A
Node 1
Node 2
Node 3
Node 4
Node 5
Zone B
Node 1
Node 2
Node 3
Node 4
Node 5

13. • OpenStack Swift and
Amazon S3 are
eventually consistent
platforms
• CAP Theorem: These
platforms follow
availability and partition
tolerance dropping
consistency
Data Consistency Processes

14. • Swifts eventual consistency
is not a good fit for
applications needing ACID
guarantees such as:
• Baking systems that must
ensure two transactions don't
withdraw the same money
• Ticketing systems that must
ensure two people never buy
the same ticket
Not a good fit

15. Common Swift Use Cases
15
Scientific /
Research
Data
Rich Digital
Media
Archive and
Backups
• Manage rich
media
content’s
explosive
growth and file
size
• Consistent
performance
at scale
• Access to a
single system
that manages
billions and
petabytes of
recordsobjects
• Object metadata
makes it
possible to
conduct complex
searches
• Legal and
governance
requirements
• Monetization
of data
• Cheep and
deep

16. Key Considerations
Performance
• Separate Account and Container
databases to SSD backed storage
• Add and remove storage
inclemently as these operations
will incur ring rebalancing
Availability
• Use a load balancer in front of the
Proxy nodes for HA
• Consult your storage vendor when
using tradition NASSAN storage
as RAID technologies will overlap
with OpenStack’s replica
technology

17. Private | Public | Hybrid Swift
17

18. 18
• Combining eplexity’s OpenStack
Swift expertise…
• OpenStack-focused, deployment
specialists
• Over 40PB+ systems management
• … with HOSTING’s World-Class
Infrastructure
• 6 privately-owned and managed
datacenters
• Specialists in business-critical
application availability
• Unique Support Experience
1
Our OpenStack Swift Solution
• No DIY required – Get started
and scale!
• Designed from the ground up for
Private Cloud

19. 19
Q&A
John Clendennen | CTO, eplexity
Mike McCracken | Director, Professional Services, HOSTING
For more information about solutions by HOSTING, please contact our
team at 888.894.4678.