Storage-area network complexity has been with us for a while now. After all, whether the SAN runs over Fibre Channel or iSCSI, the mixture of host bus adapters, switches, storage subsystems, network-attached storage heads, management-specific appliances, disks and LUNS is, in many respects, about as complicated as things can get.
Given the above, it is a small wonder that most IT managers still get by using a very limited tool set when it comes to analyzing the individual components and the inter-relationship of the various apparatus on their SANs. This is all the more surprising because such tools have been in use for years by the vendors who provide all those complex SAN products.
It is quite clear that what a storage manager needs most is a single, correlated view of all traffic across the storage network. Unfortunately, storage resource management tools are typically vendor-specific, or at least have understandably greater competence when it comes to monitoring and managing their own products than they do when working with products from other vendors. As a result, when we buy a storage-resource management solution we wind up with a useful management or monitoring capability for a vendor's own products. But we have a much lesser set of capabilities when it comes to managing products from other vendors and little, if any capability, when it is time to provide understanding of events that occur when the various assets or services interoperate with one another.
The result is that when we look across multiple storage systems and switches it is difficult to monitor performance across the different vendors' products. Furthermore, when we try to aggregate the knowledge from multiple tools we find that incompatible statistics make comparison of metrics problematic at best.
As IT shops become more complex in their operations, they also become more sophisticated in their needs. It makes good sense for many of them to start to use some of the same tools their vendors are using.
For example, the same set of probes their vendors use to test the various aspects of a test bed will be just as well suited to commercial IT environments as they are to the vendors' labs. The only difference will be that some aspects of the user interfaces will probably have to be reworked to better suit the needs of the IT community.
If you go out to look for such tools, what should you look for? Obviously, keep in mind your goal, which is to improve SAN reliability and efficiency, and to minimize downtime. This means keeping an eye out for the following:
- Early detection and analysis of device and process degradation.
- Accurate accounting of SAN bandwidth utilization.
- Accurate tracking of service levels.
- Accurate performance tuning and capacity planning.
Who provides tools that offer this? I suggest you check with your own engineering teams to find out what they use. If you don't have access to such a group, or simply don't want to be seen in public associating with engineers of an R&D persuasion, you might find it fruitful to check on the probes available in Finisar's ( http://www.finisar.com/ ) SAN and LAN performance tools, or the probes that NetScout ( http://www.netscout.com/ ) provides.