This past week I have gotten a fair amount of letters that suggest I was wearing rose-colored lenses while writing my iSCSI package. Oddly enough, Im usually seen as more of a pessimist than an optimist in the eyes of most of the people I work with.
That said, I will focus this column on some of the limitations of iSCSI to give a broader picture of the marketplace as it stands now.
First of all, the issue of performance sticks out like a sore thumb (as we saw in our tests). iSCSI is slower than Fibre Channel, no ifs, ands or buts.
10GB Ethernet is supposed to be around the corner, but I dont know of any storage routers that can handle 1G-bps at wire speed, let alone 10 gigs.
Furthermore, no native iSCSI storage is yet available. When IBM killed its iSCSI-based RAID unit (the IBM TotalStorage IP storage 200i), this essentially eliminated the possibility of pure iSCSI SANs (iSCSI from servers to storage).
As it stands now, any iSCSI implementation will be a hybrid (Fibre Channel storage to iSCSI hosts). This hybrid status could limit the management benefit of iSCSI because network administrators will still need to learn some basic things about Fibre Channel (like World Wide Names, zoning and LUN masking).
The need for security implementations in iSCSI will add to the bottom line as well, and the price of adding VPNs and other security technologies could be high.
IP networks almost always include a little bit of danger since they can be hit with denial-of-service attacks, and it is relatively easy for a person of ill will to download tools that can compromise security.
Fibre Channel has been able to get away with not having a lot of security since it operates in separate networks where it isnt as susceptible to sniffing and malicious data corruption.
Overall, iSCSI will expand the SAN market—but it wont cut into high-end Fibre Channel implementations.
Senior Analyst Henry Baltazar can be reached at firstname.lastname@example.org.