Hello Don,
> I guess that puts Wolf Mountain in the early stages.
Well, there is alot going on, but we have not announced a date of "product availability" ...
> Could you give an idea how Wolf Mountain compares to clustering
> technology already on the market?
I don't have any direct comparison information, however there are several features that we talk about that are pretty powerful concepts.
1. It is all based on Intel SHVS (Standard High-Volume Server) motherboards. This means commodity platforms. No custom hardware.
2. The clustering interconnect (which is very key to cluster cost, performance, and overall operation) is designed around our CICP (Cluster InterConnect Protocol) which allows us to support the widest range of clustering hardware. You can think of this just like our support for modular LAN and Disk drivers (supporting Ethernet, Token-Ring, FDDI, ATM, etc. on LAN, and ISA, IDE, SCSI, Fibre Channel, SSA on Disk ...) We can demonstrate clustering over inexpensive 100Mb Ethernet, ServerNet, ATM, etc. This means that customers don't have to invest in expensive technology, but can purchase clustering hardware based on their needs ... and upgrade later. This also means that with technologies like ATM we can cluster over wide area networks (that have enough bandwidth!) for distributed fault-tolerance.
> From the press releases I've read it sounds like pretty hot stuff.
It is. Microsoft is the other player that you hear alot about, but they are doing a simple SCSI based (read this as *VERY* close together) two node fail-over. It's the other vendors that are trying to do real clustering. I hear that their SCSI implementation will not even support the Compaq RAID drives!
> How big a market already exists for this type of technology?
This is a good question. I know of several key customers that are very interested right now. But most of it is for planning purposes. There are places where a cluster running the right applications could have tremendous value ... but I think many people will be 6-12 months from true pilots of these commodity clusters.
Scott C. Lemon |
