 |
|
|
| > |
QUESTION POSED ON: 02 January 2007
What iSCSI weaknesses should you consider when developing iSCSI services offerings?
|
|
|
To continue reading for free, register below or login
To read more you must become a member of SearchStorageChannel.com
With all the hype and promise of iSCSI, there are still some caveats worth mentioning.
- IP networks often have higher latency than localized storage area networks (SANs). Fibre Channel (FC) networks were designed with a simple protocol and standards limit SAN size. This means that FC network latency is often measured in microseconds rather than milliseconds.
- We have more experience with FC SANs. iSCSI SAN storage is 5-10 years newer to market. It will have some growing pains.
- Mixing normal IP traffic with iSCSI storage traffic will degrade performance. In a nonprioritized or dedicated network, iSCSI packets must compete with email, Web and application traffic. The sensitive nature of block I/O requires that it be routed to its destination with the highest priority.
- Some operating systems still don't have native iSCSI SAN storage support. It is a cruel fact that legacy operating systems are still around. You can easily find NT 4.0, Novell Netware, AIX 4.3, Solaris 2.6 and HP/UX10.20 systems running important applications. None have iSCSI support.
- IP NICs put extra burden on server CPUs. In the world of 4 GHz processors, this may not be a big deal, but without TOE (TCP offload engines), the server CPU does all the work. On a Sun SPARC server, you will see 1-2 CPUs at 100% for each fully utilized Gigabit Ethernet port. FC host bus adapters (HBAs) do all that work for the server.
- iSCSI storage management tools are underdeveloped. This is really a function of iSCSI's newness to the market. Everybody is selling a storage resource management (SRM) tool to manage SAN storage, but few address iSCSI.
|
|
|
|
|
|
|
|
Search and Browse the Expert Answer Center
Search and browse more than 25,000 question and
answer pairs from more than 250 TechTarget industry experts.
|
|
|
|
|
|
|
|
|
|
|
|
