My customer began with an IBM Fibre Channel connected to an IBM disk array that outgrew its storage capacity in less than a year (18 Gb to 90 Gb in 6 months and to 364 Gb six months later).
Now, he wants to preserve his current investment and increase his storage capacity to calculate growth for three years. I think SAN is the answer but cost is a concern. IBM's position is "throw away what you've got and install a Fibre Channel HUB and a FAStT200 storage server." What would you suggest in order not to throw away installed HW and allow for gradual storage preserving initial investment? The customer thinks he will grow in three years to 2.5 Tb. If you need installed P/Ns, I may provide such.
This is a concern for many organizations out there...
The question is always "how do I recover my investment in my current storage solution by re-utilizing my direct attached storage when implementing a SAN"?
A lot of folks have decent, very fast SCSI-3 160MB per second disks in their servers but need to implement a SAN due to the management and utilization nightmare of direct attached.
Some server vendors will let you re-use you direct attached storage in their SAN solutions as long as the storage is not too old and is certified in their SAN environment. This may be one approach to take if your server vendor also happens to be your SAN vendor.
I use multiple methods to get around this problem. One way is to take a fast server with lots of free PCI slots, fill it up with SCSI adapters, and hook up as much disk as you can to it. Then cluster this server together with a similar one, using Y cables and SCSI switches on the external SCSI connections. You can terminate the SCSI busses externally, so you can remove hosts that share the busses without taking the disks offline. Use clustering like MSCS, to share out all the storage as NAS to your clients. This makes for a real good way to re-use the disks, makes it highly available and can help you consolidate file/print servers into a more highly available environment while recouping costs.
There are also other software-based solutions that can be added to those servers and the client servers accessing them that can provide for clock-based access to the storage over IP. "Virtual Replicator" from Compaq can do this for you. It also lets you take metadata based "snapshots" of your LUNs. This is an NT only solution though.
You can also look at the in-band virtualization providers (the folks who sell virtualization appliances) as most of them use industry standard servers to run their engines and some sell the virtualization software itself. They may let you connect to the older storage through direct SCSI attach, or through a Fibre-to-SCSI bridge. These solutions can also implement IP based remote copy to a similar device at your disaster recovery site.
In the end though, especially if the current storage disks are not as reliable as they should be, is to use those disks for your LAB environment, test environment or Web environment and buy all new Fibre-based SAN disks from a reputable SAN storage vendor.
Editor's note: Do you agree with this expert's response? If you have more to share, post it in our Storage Networking discussion forum.
Dig Deeper on Data storage strategy
Related Q&A from Christopher Poelker
SAN expert Chris Poelker compares connecting a SAN with wavelength cabling and dark fiber and discusses the pros and cons of each. Continue Reading
SAN expert Chris Poelker discusses how to change the size of a LUN in a Microsoft cluster server environment. Continue Reading
Storage expert Chris Poelker outlines WWN basics in order to answer the question: "Why do HBAs in a SAN have same base?" Continue Reading