« Linux Lotus Client Continues to Draw Stares | Main| Microsoft Sharepoint Is At The Height of Success »

Build a Gold Standard Domino Server at Your Site

Category Administration
It's faint praise, to be sure, but Domino servers can run under the most bizarre configurations and resource weak environments. At the last DC Lotus Notes User Group I raised the topic of sponsoring a "Gold" standard for Domino configurations.

Here's the pitch-- at many companies I visit, and with the many administrators that I talk to, it's obvious that a surprising percentage of Domino servers are either over-taxed or mis-configured. You and I know that eventually these servers will degrade from merely limping along to a state of crash and burn. I can chart out the progress, and extrapolate a failure from captured statistics, but senior management rarely listens. The response is usually something like, "hard drives are cheap, we'll just buy more."

As I see it, many IT shops are so focused on reducing the cost of IT infrastructure, that they actually tolerate the instability which accompanies their lean budgets. After all, the IT staff is salaried, and they can put in the necessary long hours to keep operations running.

The idea of a Gold standard, is to reserve one production Domino server in every shop to be correctly configured, monitored, and given the recommended resources. With the Gold Domino server providing a baseline for stability and responsiveness, it's an easier sell to upper management to convert the remaining Domino servers to a Gold standard.

The DCNUG.org will be sponsoring a wiki for documenting the Gold standard in Domino implementations. In the meantime, the Domino 7 Performance Tuning - Best Practices to Get the Most Out of Your Domino Infrastructure has just received the latest draft update.


Comments

Gravatar Image4 - Thanks, Samuel, I've seen similar consequences of not training a Domino admin.


Gravatar Image3 - I think it's a fantastic idea.

I know a few friends whose employers ditched Notes for Exchange because of performance problems and stability. After hearing the symptoms, it became clear that no one ever bothered to do any periodic tuning, and they didn't use any policies, so settings were just tweaked ad-hoc, and probably by a few generations of sys-admins. So they dumped for Microsoft and were ecstatic. It works great!!!

Well, no @#%^! You ripped out the whole infrastructure and started all over again at 7 times of what they had to pay to build the infrastructure in the first place. The expense of having a few techs pour over a few tuning guides and spend a weekend tightening the screws would've been trivial.

Ahhh management!

You know, it's funny. Growing up, I believed this whole myth about corporate efficiency. What a crock. Most corporations are a giant cubicle hive of hemorrhaging waste.

Anyway, cheers. You have my vote.

Gravatar Image2 - I'm not really expecting that everyone is going to have to set up a benchmark server, because, as you point out, we just don't live in a perfect world. I'm looking for another way to present to management that the server with <5% free disk space is unstable because it is misconfigured/overloaded. Having a Gold standard gives the administrator a baseline which provides a demonstrative argument that doing-it-right might actually be cheaper.

I don't think a high standard means low-usage, but I think it means not being over-taxed and includes configurations like transaction logging backups, DDM, policy/settings, etc.

It's just a weapon in the rhetoric wars on justifying expenditures.

Gravatar Image1 - I think the idea of establishing a "gold standard" has merit but I'm not sure that rolling one out in every organization is going to be very successful. Just like water seeking it's own level, upgrading one server to the gold standard is just going to provide a platform to offload the work of the other overtaxed servers. I certainly wouldn't want to have to be the one to tell management that I am not going to put an application that is crashing from lack of resources onto the new server because it needs to remain in an ideal configuration.

If successful, I think it also probably opens a can of worms (that will get released soon anyway). I would hate to be the administrator that convinces an organization to upgrade all of the servers to the new "gold standard" only to find them woefully inadequate next year when Hannover is released. Maybe I am too pessimistic, but if my experience with WCS is applicable, then Hannover will probably take 2-3 times the resources required by Domino to do the same "work".
.
I still like the idea of creating the standard that establishes a benchmark for the ideal configuration since there is not enough real world information out there. (Too bad “PM for Domino” didn’t work out – by now it would have collected quite a quantity of performance information)

Post A Comment

:-D:-o:-p:-x:-(:-):-\:angry::cool::cry::emb::grin::huh::laugh::rolleyes:;-)