Analysts are unanimous when it comes to pre-purchase planning -- there is no substitute for "doing your homework". It's essential to know your current storage architecture, and have a thorough understanding of your enterprise storage needs before calling a single vendor. With this background knowledge in place, it's possible to establish a meaningful set of needs and specifications that form the foundation of a selection process. Not only must hardware needs be addressed, it is also vital to evaluate the unique software and service needs of your organization. Analysts offer their sage advice to help smooth the purchase process:
Keep compatibility and integration considerations in sight. Even the most powerful and cost-effective storage solutions are useless if you can't get them to work with your existing infrastructure. Knowing your infrastructure (e.g., SAN interconnections) and supporting software will make the selection process easier. "What's your specific storage ecosystem in terms of storage devices, storage arrays, tape libraries, storage management software, backup tools, etc.," Schulz asks. "And how well would new storage technologies integrate seamlessly and non-disruptively into that environment?" In some cases, a firmware and software upgrade may be needed to accommodate a new storage system, adding labor and potential purchase/license costs to the total acquisition. When discussing array capacity, be sure to distinguish between total (raw) capacity and useable capacity. Some vendors quote total capacity, and users find that the useable capacity may be as little as half of the total figure.
Don't let the vendor change your specifications. Whether upgrading an existing array or adding a new array to the data center, it's important to know your needs and stick to them. Vendors often tend to discuss their product strengths rather than your unique needs, so it's important to keep the conversation on track. "If you're not interested in replication, don't even let them bring it up," says Stephen Foskett, director of strategy services at GlassHouse Technologies Inc. "Don't let the vendor in there just going on and on about [maximum theoretical] throughput." Analysts also say that there's nothing wrong with "getting technical" -- invite the vendor along with one of their engineers and get their insight.
Standardize on a vendor for each tier. As more products are introduced to the data center, the IT staff is obligated to learn and support additional products. Although analysts recommend considering every suitable candidate (including small vendors), there's a need to balance the number of manufacturers represented in the organization. "It wouldn't make sense to have two vendors on the floor at the same tier of storage," Foskett says. Instead, Foskett suggests that it may be more effective to go with one vendor for each different class of storage. This strategy allows the most competitive vendors to play a role in your storage objectives without necessarily overwhelming an IT staff.
Buy the best storage array that you can afford. This may seem like a serious budget issue, but analysts are quick to note that storage demands are increasing rapidly, and organizations are often better served by investing additional resources to acquire the best possible product up front, because it will offer you much more scalability and flexibility into the future (this is where an understanding of corporate storage needs and future expectations is important). "It's better to spend a little more money on a lightly-used enterprise-class array than it is to save some money and end up with an over-used mid-range array," Foskett says. It may be possible to get more value for your storage investment by timing your acquisition to the last month of a vendor's quarter, or the last month of the vendor's fiscal year -- typical "sale periods" where it may be easier to negotiate a better price or obtain attractive upgrades.
Consider iSCSI for mid-range applications. Analysts suggest that iSCSI may finally be an appropriate solution for applications where low-end Fibre Channel installations may otherwise have been used. "If somebody is in a position where they can't quite afford Fibre Channel, then they should go with the iSCSI," Foskett says. He notes that iSCSI array pricing is very attractive (especially for Windows platforms), and the associated multipathing and replication software is sometimes included with the iSCSI array -- resulting in a substantial cost savings in software as well as hardware.
Get a handle on the "effective price". There are more costs to a storage array purchase than simply acquiring the array itself. Even when one product may have a lower price tag, it may ultimately prove more costly. "The [total] effective price may come in lower when all the pieces are finally added up," Schulz says. Such pieces may include shipping, freight, install costs, facilities preparation, paying movers, data center cooling updates, software purchases/licensing, warranties, service agreements, staff training and provisioning/migration efforts. Unexpected expenses can add substantially to the final array cost.
Go to the next part of this article: Storage arrays: Avoiding integration problems
Or skip to the section of interest: