Typical Uptime - AlphaSever/Stations

[DaveGPearson]

This is a little bit of an open request but for the installation and application I'm trying to justify it would really help me if I could get some 'real' uptime data from the wider industry on how reliable AlphaServers are. I know they have outstanding performance and are used by half of wall street etc. but it would be great to know from the real guys who have to keep these things running the real facts behind the claims.
The two models I'm interested in are Alphaserver DS25 (a number of, both single and dual processor) and an Alphastation DS15. All will be running Tru64 Unix.
Simple feedback such as "Typical application, date installed, MTBF/uptime/failure rate etc., whether frequently power-cycled or running 24/7" would be great. Direct contact details to confirm this would also be appreciated but not essential.
The reason for this request is to support the use of the Alphas in a mission-critical role running a legacy app as part of the business justification.
Posting to the forum preferred but mail to hp@pearson.in also accepted if confidentiality required.

[DaveGPearson]

I should have added that it is Tru64 Unix 5.1B

[Vladimir Fabecic]

Hello Dave
Not sure if I understood the question.
If so please supply more details.
From my experience DS25 machines are quite stable. One of my customers has several DS25 servers (single processor) running Oracle database. They are switched off once a year for maintenance.
Failure rate?
Few disk failures but it should be no problem if you have RAID controller.
So from my point of view, very stable hardware.
If you plan to run mission critical applications, think about TRU cluster.
Have only two DS15 machines which are test servers for developers, so they are rebooted more frequently. They are working for more than two years without failure.

[Rob Leadbeater · Accepted solution]

Hi Dave,

I used to manage 3 DS25s running Tru64 5.1B and TruCluster which never had a hardware failure in 3+ years. That said, they had no local disks - everything was on the SAN.

These were test and dev boxes, so they were rebooted fairly frequently.

Servers were running Oracle 10gR1 database, which was much less stable. It was this that warranted the servers being rebooted so frequently. Your uptime, will depend entirely on how well the application is written, which you've given us no information about...

Whilst Tru64 5.1B is very stable, I would have to question the logic of using it in a new installation, even if it is supporting a legacy application...

1. You can no longer buy the hardware new. HP stopped taking orders in May 2007.
2. Your support costs will be high, compared to other hardware/UNIXes.

Cheers,

Rob

rob [at) leadbeater [dot) org [dot) uk

[DaveGPearson]

Thanks for your responses.
To clarify the situation, we are deploying the Alphas in a Safety-related system to provide HMI and Alarm Logging capabilities.
The system is being spec'ed this way tto take account of the development platform used, and previous configurations in similar plants elsewhere.
In any other application, use of Itaniums etc. would be fine but we have a large burden of proof we are relying on on the existing platform.
There is a vast degree of anecdotal evidence as to the reliability of Alphas but no empirical "hard" data to back this up (or I'm not aware of it). Hence the plea for the reliability data. I know a lot of unreliability is down to the application layer playing up but if we know of (for example) flaky supplies, error-prone memory, inconsistent build quality etc. that can work against our "high reliability" claims then this all helps the cause.

[Steven Schweda]

> [...] flaky supplies, error-prone memory,
> inconsistent build quality etc.

I tend to buy only used, obsolete junk, but
DEC/Compaq/HP get their power supplies from
the same vendors as everyone else, every
Alpha system I've ever seen uses ECC memory
(so memory errors are hardly a worry), and
I don't know how I'd recognize "inconsistent
build quality". All my junk looks well-built
to me.

If continuous availability is the goal, I'd
probably be looking more at building clusters
of some kind, where a single-point failure
might be more easily mitigated. (My
experience in this is limited to VMS. I
don't know much about TruClusters, beyond the
name.) Any piece of hardware can fail, and
when it does, the probability of that event
ceases to be important.

My biggest worry about acquiring Alphas
running Tru64 would be that serious
development of Tru64 appears to have stopped
a while ago, so availability of new
applications tends to be poor and declining.
If you intend to run only existing (or
internally developed) applications, this may
not affect you.

For what it's worth, I've been running
antinode.org continuously on VMS on a used
XP1000 since 2005, and on a new (old stock)
AlphaStation 200 4/233 for about four years
before that with no memorable hardware
failures. I also had a used XP1000 which had
a power supply die under warranty, and whose
main board failed, apparently during a car
ride across town. (I'm guessing ESD, but
it's a mystery.)

[Vladimir Fabecic]

Few days ago I spoke with one of my ex customers. They migrated from TRU cluster and alpha servers to linux on itanium based servers.
They are running Oracle database.
And they said they miss stability of alpha servers and tru64. They had oracle on TRU cluster for four years. Never had unplanned downtime, never had OS related problem (patching OS and Oracle was planned).
But NO NEW hardware fact did the "job" and they had to migrate. No future for alphas and TRU64.
Sad but it is reality.

[Jan van den Ende]

Dave,

I know not much about Tru64, but since you asked about AlphaServers, here comes.
Personal experience: 4 node VMScluster with 10+ years uptime.
Single nodes had planned downtime (without affecting CLUSTER/application availability) for software AND hardware upgrades; but individual nodes have had 3+ years uptime.

From what was TAUGHT to me during Tru64 cluster courses, similar achievements can be had on Tru64. (Of course, that DOES take some carefull planning). Caveat: No personal experience, nor first-hand knowledge of such practices.

But: Alpha servers are generally VERY robust!

Success.

Proost.

Have one on me.

jpe