Major vendors reveal blades
Summary
Topics
Now, just a few months later, HP finds itself being elbowed not only by its merger partner, Compaq, but it's getting poked in the ribs by blades from Dell and soon from IBM as well.
Admittedly, HP did something atypical. Instead of developing its own proprietary design, it based its blade architecture on an open bus architecture: CompactPCI. This is a 7-year-old accepted bus standard that uses a 2mm pin and socket connector; the CompactPCI boards are inserted from the front of the chassis, with I/O exiting either through the front or rear. It's an industrial-strength bus that has found wide acceptance in a broad range of applications that require high bandwidth and HP added its own extensions to make it more than your average telco box.
Compaq's proprietary BL10e blade design (I'm going to ignore Compaq's claim that the standard Ethernet interconnect ports on the backplane disqualify it from being proprietary) puts emphasis in the details. The BL10e blades, compared to the older ProLiant DL320 line, has a larger base memory, no slots, integrated administrator functions, and a form factor that will fit 20 blade servers in a 3U enclosure. (That's 280 BL10e's in a standard 42U rack.) The current level of Compaq's blade architecture is fairly ho-hum (700MHz ultra-low voltage Pentium III with a 100MHz front side bus, 512MB (expandable to 1GB) of PC133 ECC SDRAM, and a non-hot pluggable, 30GB ATA/100 4200rpm hard drive--which is also the maximum drive capacity).
While the competition was duking it out with single processor blades, Dell slid in under the radar with the PowerEdge 1655MC, a dual-processor competitor capable of a maximum 146GB of internal storage, redundant Ethernet paths into and out of the enclosure (among other neat stuff) and some interesting management tools. Unlike Compaq's design, there's little trade-off in memory or storage to achieve density. The downside to its versatility is that only six blades (up to 12 CPUs) will fit in a 3U enclosure.
For all the blades that already exist, however, IBM's is the wild card here for the near future. It has been offering RLX-based blades for quite a while, but its "high function" blade servers won't debut until Q302. IBM is being fairly tight-lipped about the particulars, but it's not difficult to speculate that it could offer solutions based on Intel or IBM processors, or both. If IBM marketed a blade with a single CPU using its Power4 shared-core dual processor, it would actually be providing 2x the potential performance of its rivals with half the software licensing fees attached to 2 physical CPUs. Any multiple of one CPU per blade would, likewise, increase performance potential while minimizing per-CPU fees. On the Intel side, IBM has already been working with Pentium 4 Xeons in a variety of ways and, while they're not currently low-power processors, there's ample evidence that more than a few of you out there prefer punch over power savings
Blades are still an emerging technology, making everything subject to revision. Personally, I'd keep a good eye on IBM. It out-gunned Sun on Q401 worldwide Unix sales, and dunked Dell during that same period in high-end 4- and 8-way Intel products. While initially trying to lure folks away from the small server side of business, there's sure to be a commotion now that it's turned its attention to the market.
Do you think one vendor's blade design will prevail? Which one, and why? Share your thoughts in our TalkBack forum, or send an e-mail to Bill.
Talkback - Tell Us What You Think
The best of ZDNet, delivered
ZDNet Newsletters
Get the best of ZDNet delivered straight to your inbox
