In-Depth
Q&A: The Optimized Data Services Model
If you need to provision, protect, migrate, dedupe, encrypt, replicate, recover, and archive data sources, ODS may be just what you’re looking for.
If you want to simplify how you provision, protect, migrate, dedupe, encrypt, replicate, recover, and archive data sources, ODS may be just what you’re looking for. To learn what ODS is and how it works, the benefits you can expect, and how to avoid common ODS project mistakes, we spoke with Chris Poelker, vice president of enterprise solutions at FalconStor Software.
Enterprise Strategies: What do you mean by optimized data services?
Chris Poelker: Data services are the four aspects of what an IT department does to enable the applications running the business: storage provisioning, data protection, data replication, and application recovery.
An ODS utility is created by virtualizing existing data, storage, and servers where possible to enable physical abstraction and flexible data movement between compute and storage elements. Once virtualized, the ODS platform or components should allow the creation of policies that enforce specific service levels for explicit or pooled datasets. The grouping of data elements for consistency or recovery purposes should not be hampered by physical constraints (i.e., LUNS in the same array, or SAN vs. non-SAN or storage-network-attached devices or hosts).
The overarching goal of the ODS model is to optimize and make more cost effective and operationally efficient (i.e., simple) the ability to provision, protect, migrate, dedupe, encrypt, replicate, recover, and archive any data source to any application in real-time via policy.
What benefit does ODS promise?
The cost benefits can be quite dramatic, and in these uncertain times, that’s a good thing. The main benefits of an ODS model is the ability to get better return on current investments by leveraging existing assets and making them more efficient. Another benefit is the commoditization of storage through virtualization, which also enables complete data mobility. Imagine being able to eliminate all of the array-based licenses across all the storage in the data center.
Since provisioning and protection in the ODS model occurs at the fabric layer, operations can typically operate much more efficiently, and since the model calls for re-use of much of the existing policies currently in place, the learning curve is reduced. One goal of the ODS model is to simplify application recovery by negating the requirement to map out all the intricate application inter-dependencies for recovery and to provide better service-level-agreement (SLA) standards across all applications more efficiently at a greatly reduced cost than the current standards.
There should be no requirement for new budgets to implement an ODS model, since implementation should be based on the savings it provides.
Why isn’t leveraging the ITIL standards (and reaping the benefits they provide) sufficient?
One of the great benefits of ITIL is the reduction of IT costs via standardization, but ITIL alone is not enough anymore. The ODS model goes a little further than ITIL to reduce IT expenditures even further while providing much higher service levels for applications at a significantly lower cost than what I am calling “traditional methodologies.” These traditional methodologies include (but aren’t limited to) array based provisioning, the backup process, current approaches to data replication, and disaster recovery techniques.
Where can you apply the ODS model, and what kinds of technologies fit into this model?
The four areas where implementing an ODS model can provide immediate benefits are storage abstraction with thin provisioning, data protection, data replication, and application-recovery techniques. Familiar technologies such as virtualization are used here, but non-technology components, such as setting policies, are also a major aspect of the model. A simple explanation of the ODS model is “the implementation of an intelligent abstraction layer over existing physical resources, where policies are created to optimize all data services.” As an example, for provisioning, all data would be brought under the abstraction layer in place, and storage resources for new data would be provisioned from pools of storage with known performance and reliability metrics.
On the data protection front, backup and all the data movement implicit to that process would disappear, and protection would be either continuous or periodic based on the SLA policies provided for the applications in question.
For data replication, array-based replication could be replaced by continuous fabric-level replication so that all applications and platforms could use the same solution operationally. Since mission-critical applications require immediate recovery, hashed-based data deduplication of the datasets cannot be used. Avoiding duplication via sector- based micro scanning of application data provides better efficiencies for moving data across the network while eliminating the need to reconstitute the data on the other side. Traditional data deduplication would be leveraged for non-mission-critical applications to provide efficiency in data movement for those applications. Since all intelligence for these operations would occur at the fabric level, and the ODS solution would optimize existing IP networks for data movement, costs would be dramatically reduced, while recovery objectives (RTO, RPO) would radically improve.
Since all applications would be continually protected and the ODS solution integrated within the application consoles themselves for recovery, all recovery would be holistic; and from the application’s viewpoint, administrators would be able to leverage the interfaces they are familiar with and recovery would be simple and immediate.
In tight economic times, leveraging existing assets (rather than investing in new assets) must certainly be attractive to upper management, so it shouldn’t be difficult to get management’s buy-in to such projects.
That’s true, but a project of this scope needs to have the focus and attention of upper management and be driven from the top down to provide political oversight. Implementation of an ODS model may ruffle some feathers as some long-standing relationships between incumbent vendors and IT managers are either reduced or eliminated.
What mistakes does IT traditionally make getting started with ODS?
Since the benefits can be dramatic, one mistake is to try and “boil the ocean” by implementing all aspects of the ODS model all at once. A better way is to take a phased approach and introduce the model step by step, starting with fixing backup or disaster recovery, which is usually where the company is feeling the most pain.
What best practices can you suggest IT follow to avoid such problems and maximize the benefits of ODS?
In order to adhere to the ODS model and not waste money, IT managers should ask themselves these five questions prior to implementing any new technology, whether in the data center or at remote locations:
- Does the solution simplify operations?
- Can we use the same solution across all platforms and applications?
- Does the solution leverage existing assets?
- Can we leverage current policies and procedures?
- Can we implement the solution based on the savings it provides rather than relying on new budgets?
What products or services does FalconStor offer for ODS?
FalconStor is a driving force behind simplifying the data center and reducing IT costs for our customers. We have embraced the ODS model through our data services platform known as IPStor, from which we create a family complete solutions based on the ODS model: FalconStor Network Storage Server, FalconStor Continuous Data Protector, FalconStor Virtual Tape Library with Deduplication, and FalconStor’s file-interface deduplication system (which is currently in beta).
Unlike most technology vendors (that only offer a single solution or that buy up companies and tie everything together with scripts), we can leverage our platform to reduce complexity -- and we can call each one of these interrelated solutions a “platform for optimized data services,” or PODS. The PODS provides all the ODS functionality requirements via a single console.