Early report: New 3000s satisfy, disappoint

Early report: New 3000s satisfy, disappoint

Summer bakes early responses to a trickle in A-Class, N-Class study

Was it the weather?

We’re not sure why, but responses to our survey on satisfaction with the new A-Class and N-Class e3000s were as scarce as rain in a Texas summer. We asked in August — the most vacation-ridden month of the year — and we know that plenty of A and N deliveries are backed up, waiting on multiprocessing abilities that are only available with the 7.0 Express 1 release shipping this month. Maybe a string of summery days extended the “out of the office” replies we received.

Whatever the reason, we promised to report this month on what we heard from our readers about their new system experiences and how their expectations have been met. In general, the sharpest comments came from customers who decided to pass on purchasing an A-Class system, especially those in the development community. Those most critical of the value proposition in the new e3000 low-end were those with the longest tenure in the community, who know the systems’ technical capabilities best. Many disagreed with HP’s choice to limit the processor speed on the A-Class systems (140 MHz for an A500, versus 440 MHz in the HP 9000 A-Class systems). But most customers who bought an A-Class are satisfied, and those buying N-Class machines seemed to fall into an even happier category.

We gave respondents the option of replying anonymously, and many took us up on the offer. Few of them with negative comments wanted their names on the record, perhaps a sign of their long-term commitment to the 3000 community. We’ll keep asking through the fall, while HP ramps up the shipment of the new systems, and report back later with more complete results.

Our A-Class has been a huge disappointment. HP didn’t have the processor configuration we wanted at the price we wanted. I have to stay with the new technology for our customers’ sake, but we’re trying hard not to roll any A-Class systems to our customers. We wouldn’t recommend the box for a higher load factor.

We acquired a 220-MHz N4000 system to replace an 988 that was on lease. The N-Class was chosen because it was faster and less expensive than the 988, and the available A-class (single-processor) would have been a step backward. Although we have not measured the performance improvement, it is significantly better than the 988. Activities like full and partial backups are almost twice as fast (probably a combination of processor, PCI bus and DDS-3 tape drive upgrades).

We have found a few third-party applications that are experiencing some problems (MPE 7.0 is suspected). Overall we are very satisfied with the new equipment. The lease cost (with three years of hardware and software support included) is substantially less than what we were paying for the 988.

We have two A-Class boxes on order, but none in-house yet. We are replacing a 960 and a 980 with these boxes. We chose A over N because of price — especially with third-party licenses. The marketing team at Cognos won’t be benefiting as much by our choice of the A-class. We didn’t need a big jump in connectivity or CPU power on the systems we’re upgrading, so the N-Class was unnecessary for us.

The A-Class is priced very well, we think. However, HP’s limiting the processor speed for MPE/iX makes them look pretty cheap. How many more As would they sell if the CPU power was not held back?

Our A500-100 (2Gb memory, 72Gb disk) was installed about a month ago. We ordered the A500-200 but we have to wait until Express 1 comes out before HP will ship and install the second processor. We put it into production at the end of July. It is replacing our seven and a half year old 937LX (64 users, 128Mb, 4Gb, MPE 6.0, Image).

According to the performance charts, this box should be 2.5 times the power of the 937. We’re experiencing performance increases more in the 3 to 4 times range. We’re very satisfied. Of course, the best thing for us right now is having no user limits. We were bumping the 64-user limit regularly (we had sessions limited to 60). I now see up to 70 sessions active and the system loafs along at about 15 percent usage.

We have no need for N-Class power (or upgrade/support expense). The A400 just didn’t seem like enough of a power increase to be worthwhile. The A500-100 seemed fine, but for just the cost of the second processor, ordering the A500-200 was a no-brainer. The second processor didn’t change upgrade fees or support costs from any vendor except one (for our ODBC product).

I don’t know about the price. It cost about the same as the 937 did nearly eight years ago. That’s not that bad. Price-wise I was much more concerned with Cognos than HP. With Cognos “only” sticking us $25,000 to upgrade, I guess I felt it could have been much worse. Support costs will increase 30 percent-plus over the 937, but it looks like our processing power will increase five to six times.

We were running out of gas on the 937. I really was hoping not to have to upgrade the 937 at all since (three or four years ago) I thought we would have transitioned from our HP 3000, custom PowerHouse code based environment to a more mainstream NT network, “packaged software” environment. Things are starting that way, but I can easily see the 3000 going on for another five years.

HP killing the 9x7 line forced our hand to do this now, but with the anticipated business increase we’re expecting next year, we would have been forced to do some kind of upgrade anyway. I’ve kept expecting the workload on the 3000 to decline over the last few years, but every year it increases! The 3000 is still business-critical for us, running not only the company here, but our sister company in Utah as well.

We don’t have an A-Class, but did some testing on one. It was about 40 percent faster doing serial disk reads than a 968 (64 MHz). I expected a lot more (newer disk drive, better IO capabilities). I wanted a 440-MHz A-Class at the 110-MHz A-Class price, or less.

We’re waiting for a machine that is price/performance competitive with the low-end HP 9000 systems (like the HP 9000 A-Class), or with Sun’s low-end Blade (that’s a 500-MHz 64-bit RISC system for $950 with 128 MB RAM, 15 GB disk, and CD-ROM).

HP’s own positioning of the HP 3000 A-Class emphasizes the need for performance (e.g., Web server, Internet-accessible databases.) Yet they decided to drastically limit the capabilities at the low end.

We’ve been looking at the A-Class for a while now. We’d be replacing a 957 and a 977. Our problem is that we have to continue to use HP Cooperative Services, and no one can tell us for sure whether it will work. We don’t want support for Cooperative Services — we just don’t want to spend $250,000 for a machine and then find out we can’t use it.

I have installed six N-Class machines and two A-Class machines for customers, some new, some replacements. The customers have been very happy. For example, one recent customer had a job that went from 45 minutes to five. The install goes pretty well. One complaint is the same one that those that are using MPE/iX 6.5 have: STM stinks. Hard to use, little documentation.
There are some little problems that I hope will be fixed in Express 1. For example, when taking down the system it usually hangs at Shut 4, so we have to go away for five minutes.

The biggest complaint I have heard is from upgraders. Third-party software companies that I know of are asking for lots of dollars. And the customers are only changing hardware because they have to. They aren’t after greater capabilities or enhancements. I know of three customers that have dropped products rather than pay the upgrade price.

Like several other HP 3000 vendors I know, we couldn’t bring ourselves to buy a crippled box (A-Class) — even with the CSPP developer discounts. Instead we did like most of the others — picked up a small 9x8 box to do our MPE/iX 7.0 testing on. A shame. If the price was just a little better and the box wasn’t crippled, we might have picked one up. It would have been our first “new from HP” HP 3000, too.
While the N-Class seems to be a good deal, and I know of a few sites that have bought them, the A-Class seems to be a poorly executed idea to me. HP continues to paint the 3000s into a corner.

We are very satisfied with our new N-Class box. It is running about three times as fast as the 959/200 that it replaced. We chose the N-Class because of our size. We are currently running about 350 users and 15 jobs. We are planning to expand to 500 users by 10/01, then to approximately 700 users by 2003. With our 959/200, we had a 250 user license. It would have cost about $100,000 to expand it to a unlimited license. The N-Class was easy to cost-justify.

We were first told that our 929KS-020 wouldn’t be supported after February, 2002 during our budget planning. As a result our customer decided to upgrade an N-Class server this year.
Because less than three years had passed since our last purchase, our customer has decided to migrate off the HP 3000 over the next three years. The announcement in The 3000 NewsWire that our 929 processor would be supported until January 31, 2006 made the decision to migrate even stronger in the mind of the customer. The news of late of HP entering a market, and then abandoning it within just a few years or so, has done nothing to strengthen HP’s position here.