Software-Defined Storage (SDS) may hold a lot of promise. But implementation can be far from straightforward. Organizations may have hardware and software dating back several generations. So what is the best way to go about it?
1. Match Your Servers
Mario Blandini, vice president of marketing, SwiftStack, said to choose a software product that supports the server hardware you already own today. He noted that appliances cannot be software-defined as they are defined by their fixed hardware appliance form factor.
“The full benefits can only be realized when users are free to run software on the same brand of servers they use for other workloads, with the same freedom from lock-in they enjoy in compute,” said Blandini.
2. Pay As You Grow
A follow-on tip that is valuable during the lifecycle of SDS is paying as you grow. Why? Flexibility and agility are arguably the biggest benefits of SDS, and new capacity or bandwidth can be added to SDS on the fly.
“It is best to deploy only what will be used in the near future, and continuously monitor utilization so additional hardware can be ordered and added a quarter in advance for strong utilization rather than three to five years in advance and risk going forever unutilized,” said Blandini.
3. Know Before You Go
Several experts recommended having a clear objective of what you are trying to accomplish with SDS. If your objective is it to eliminate vendor hardware lock-in, for example, make sure you are aware of potential software-defined storage vendor lock-in, too. Understand what other hardware and software dependencies and interoperability are needed for your preferred solution. How will your software-defined storage solution integrate with existing management tools and processes including data protection (backup, BC, DR) among others?
“Keep in mind that different software-defined storage solutions address various needs from bulk storage, to data protection as well as general purpose, so understand the options as you may be comparing apples to oranges,” said Blandini.
4. Consider Other Costs
When thinking about implementing a first-generation, software-defined storage solution, make sure to include much more than just the hardware and software costs as you calculate your total cost of ownership (TCO). Add in the extra space, power and cooling costs for a lab to implement and test a solution before putting it into production, and put in headcount costs (and expertise) needed to implement a solution.
“These are all important costs that must be factored in as a part of all new software-defined solutions,” said Steve Kenniston, vice president of product marketing, Infinidat.
5. SLA Guarantee
SDS may sound attractive, but it still involves hardware. So be careful that vendor support will meet your SLAs and overall business needs, otherwise you can get stuck in the middle of potential vendor finger-pointing. Too often, users call the hardware vendor to ask for help integrating the solution with their software, and the vendor says, “The hardware is running fine, call the software vendor.” Predictably, the software vendor then says, ‘It worked fine when we tested it.”
“Make sure that the integration and vendor support of the myriad potential hardware options for a software-defined solution will meet your SLAs,” said Kenniston.
6. Look Beyond Storage
SDS is only one aspect of the overall push toward the software-defined data center (SDDC) or software-defined everything. So it is important to look beyond the storage side.
“As you are consolidating the infrastructure, you need to see SDS as more than just deploying a new storage solution and instead consider all aspects of the environment that surround the software-defined components,” said Kate Davis, manager, HPE Storage marketing.
Questions to ask include the following: Can you refresh the servers with data in place or will that require a migration? What’s your longer-term hypervisor strategy — can you easily change down the road? Do you have data mobility options – can data seamlessly move systems, generations, tiers, locations? Is the network modernized to support added traffic?
7. Hypervisor Integration
In many ways, software-defined storage is an extension of virtualization technology. So it makes sense to consider the hypervisor carefully. To make implementation and management as easy as possible, start with a solution that is tightly integrated with the hypervisor, said Lee Caswell, vice president of products, storage and availability, VMware. “When shifting to a software-defined infrastructure, the integration between the software components, especially the hypervisor and software-defined storage layer, is key.”
Tight integration with the hypervisor makes the management experience seamless by naturally extending the skillsets and management tools used today for server virtualization. The integration also means that critical virtualization services, like dynamically moving VMs as needed, and high-availability features, natively work, providing greater functionality without sacrificing performance or additional management steps. Additionally, the tight integration can ultimately lead to a significantly lower-cost solution by enabling greater VM density per servers. With tight integration, the performance of the storage can be higher thanks to shorter I/O paths and lower overhead, delivering more of the CPU and memory resources for real workloads versus the storage overhead.
8. Review All Storage Repercussions
The software used in SDS needs to know how to address more than the storage file itself. It also needs to be aware of additional storage specifications such as: high/low latency, online or disconnected, R/W or WORM. This gets even more complicated when dealing with cloud storage, as each specification can change the unit cost dramatically.
“In today’s world where the move to cloud storage is very active, it is important for the software that drives it to be cost-aware and present the customer with the best balance between acceptable performance and cost,” said Tibi Popp, CTO, Archive360.
A sound forecast on future storage needs is essential – as is an understanding of the analytics needs of the enterprise. Many companies are figuring out how to better analyze the mountain of information they own and use it for competitive advantage. Implementing SDS without taking this into account could prove expensive in the long run.
“Start with a data strategy that includes where your data currently resides and the analytics/business intelligence requirements to leverage this data,” said Ihab Tarazi, CTO at Equinix. “Forecast the volumes of data and the performance criteria in terms of throughput and latency. This will point to one or more software platforms for analytics and storage and a deployment plan.”
10. Know Your Content
“Know thy content” very much applies to SDS implementation projects. Giving this plenty of upfront thought process and planning allows you to set up a powerful, accurate and efficient storage environment. “Measure twice, cut once,” the phrase pounded into aspiring carpenters, should be applied to SharePoint and other content. In other words, know your content and know what you are trying to accomplish with it. The initial rules you establish for content management will dictate how well the long-term solution works.
“In a more general software-defined storage solution, this translates to establishing content and management goals, building a deeper understanding of what content you will be managing, and ensuring the storage rules reflect these goals and content,” said Paul LaPorte, director of products, Metalogix.