how many sys admins does it take .... ?

[John Bolene]

we have 5 to handle around 110 HPUX, 10 AIX, and a handful of W2K servers

I would think that you need at least one more person to help.

[Paula J Frazer-Campbell]

John

I think your company has a problem - yes you have a high workload, but the problem is DR (The sysadmin/staff are a part of their DR solution), how do they manage for your holiday, sickness, courses and God forbid you having an accident?

Paula

[Radim Jarosek]

Hi,

we have :

4 HP-UX servers
16 HP-UX workstatations
4 Linux
3 Oracle databases

we are :

1 dba/sysadmin
1 sysadmin backup


As someone said, it depends on users ( how you train them ;) and how you automate daily tasks (watch scripts for system/db for space, performance ... )


HTH

Radim

[John Dvorchak]

We are a very large Telephone company and we try to keep the ratio of 10-15 UNIX boxes to each SA.

The last company I worked for had One UNIX admin for approximatley 35 Workstations and 2 Servers. Two NT admins for 25 Nt servers and 4 desktop people for 1500 desktop's running NT 4.0

Hope this helps, but at the very least you need another person in your shop. One person can not possibly be expected to be available 24x7 365 days/yr. That means no sick days, no vacation, no getting injured or killed. I know if you're dead you don't care but who ever is in charge should realize that.

If they don't get you some help, I'd update my resume if I were you.

[Martin Johnson]

12 HPUX
70 AIX
118 Solaris

9 Fulltime/ 2 part time SAs

[John Kittel]

I'm very healthy, haven't been sick for years...

Courses? What are courses? ( Oh, you mean offsite training courses! Haven't been to one in many years... no $$ for that)

I'm an atheist, so God won't protect me from accidents ( maybe would even cause me to have one?) - so that's a wash I guess.

Vacations, I think I've been called once every vacation I've had, but fortunately always very minor stuff. Except my last vacation, I managed to find a place where there are no phones, even land lines, no net access. I think if they shined a search lite with a cutout bat silhouette onto the clouds at night I could respond with smoke signals the next day.

As I mentioned there are people here who can each cover some limited areas of my usual resonsibilities, though I think there's a huge vulnerability in that no one other than me is able to restore backup tapes on several systems. Disk crashes are rare enough it hasn't happened. Last disk crash was day before my last vacation.

I've actually been told that in labor overhead cost cutting efforts, manangment doesn't really have any good way of knowing how deep to cut, so the "strategy" is to keep cutting until something really bad happens, then they will know they cut too much and can add some resources back... - hence my personal quote :-)

- John

[Pete Randall]

John,

All I can say is that your management is at least, if not more, short sighted than mine. How many hundreds of thousands of dollars do they need to lose from one catastrophic failure before they realize what cheap insurance adequate staffing is? I honestly hope that you never, ever have to go through the "management awakening experience" that they (your management) is begging for.

Good luck,


Pete

[Pete Randall]

"are" begging for.

Sorry!


Pete

[aaronleung_1]

Hello,

I am in a telecom company of hk. We have more than 120 Unix servers with all different platform (HP-UX,AIX,Solaris and Linux)

We have one Sysadmin for all these unix servers.

[aparna challagulla]

hi john,

i am working in a financial institution and we have 12 hp-ux, 2 linux, 16 windows servers and two sys admins to handle them.

we are under 24/7 support from hp as well.

we work in shifts as all the maintainance work has to be done in the middle of the night/ weekends when we get some downtime.

regards,
aparna

[steven Burgess_2]

Hi all

I think I must be fairly lucky that in my 2 IT roles the companies have been fairly large scale

Current

190 Unix servers and growing (In the UK) over 500 worldwide (we only support uk which may end soon because the parent company in the states are working in a saturated market and are trying to grab all our stuff)

Previous

Over 300 worldwide

Hundreds of databases, network links , bespoke applications etc


There are over 200 people in OPS here and at least 300 in my last place all doing various roles

From a unix perspective both companies have been split into the 1st,2nd << 24*7 ,3rd << on call and projects << travelling roles with around 6 people in each team

The current firm earn billions worldwide yet we have just been advised that there is no money for external courses we are taking more people on but even struggling to fund for a new desk in our office!!!! (bizarre).

They have secured a deal with HP for e-courses, which to be honest are ok if your starting out but when you need to strengthen your skills which to be honest I do, doesn't really help.

HTH

Steve

[Jim Mallett]

John,

There are no such things as frivolous questions, just frivolous people that ask them.

I actually have enjoyed watching this post over the course of the past couple days. I'm in my first real run as a SysAdmin and haven't had any resources on the HP end of things (other than the forum which I am eternally greatful for). It's interesting to see what resources people do and don't have.

I think you've found that there is no 'typical', that being said here is what I'm responsible for:
5 HP-UX (You mention the word Unix and people scatter.)
1 AIX
1 Really Old AT&T Conversant running SCO maybe?
4 Windows 2000 Servers
1 Symmetrix (I guess I'm the storage administrator now too?)
1 Centera

I have a backup, unfortunately the extent of his knowledge is taking the DLTs out and replacing them. (I won't even comment on how often that is done incorrectly.)

There are 3 Admins for our mainframe.
There are 4 Admins for our other 30 or so Win servers.

I know too well that 'overworked' feeling, and I have a seemingly small environment. My one wish would be to have one other experienced Unix person around, at least to have bounced ideas off while I was really green. I found the forums a bit late. What I wouldn't have given to be a Jr Admin for one of the Salad Heads.

Good luck...
Jim

[Sandro Schaer_1]

Let's see what we have here :

- 13 hp unix servers
- 3 clusters
- 2 ncr unix servers
- 50 oracle databases
- 30 different applications
- development/maintenance process management software (written by me)
- development/maintenance system management software (written by me)
- development/maintenance application software (partly designed/written by me)
- performance tuning of
- pl/sql
- forms
- cobol with embedded sql
- databases
- project manager new projects
- consulting for all the technical stuff
- support for daily business problems
- some training/education of developers (>20)
- stand-by for any problems 24 hours a day, 365 days per year (holidays with cell phone and laptop only)


That's my job. No second UNIX or Oracle admin.

Sounds like fun, doesn't it ? :-(

[Mic V.]

John,

IMO you may be overworked. Here's our info:

Computer-related mfg company with 3000+ employees and global coverage. From 1997 to 2002, in IS we had:

- 1 UNIX admin (started with 2 HP-UX servers and ended up with 25 HP-UX servers including two MC/SG clusters, 10+ Sun systems running DNS,UNIX firewall, corporate backups). There was no backup and the job involved 24x7 on-call.
- 2 part-time VMS admins (their other jobs were manager and NT administrator, respectively) for 4 production VMS clusters, approx 10 development machines.
- 1 full-time NT admin for approximately 30 servers comprising one domain (but interfacing with other domains)
- several desktop contractors
- two full-time network guys doing everything from running cable to managing the firewall

We had over 50 people in IS at the peak.

In 2002, one VMS admin left. In 2003, we made several changes: laid off the VMS/NT guy, laid off the full-time NT guy, brought in an NT/web site/UNIX guy to manage the group, and added a UNIX/router guy.

We still feel like we're facing an impossible battle to get everything done. I don't blame you at all for feeling overworked. I think this is something we're all having to face, because we've been dealing with the overworked situation since the US economy took a dive. Now management probably thinks "wow, what great productivity" and doesn't realize how we're struggling just to keep going, that we don't have anything else to give.

HTH,
Mic

[Elmar P. Kolkman]

The main problem we (as sysadmins) all face is our pride... Even though we are managing the servers with too little resources, we don't like the machines having problems, so we overwork ourselves just to keep things running. The management only sees the machines running, not the people who overwork themselves. So they think: "Hey, we're doing great... All those servers can be managed by just this little number of people.".

The only solution that might work is letting a problem really give them something to think about. Let their fileserver or mailserver crash on them and need a lot of time to repair it, or something like that. But our pride prevents that...

I've been in numerous of companies, and almost always big companies that knew they needed the people, so I've been only once been in a undermanned situation (and it really went wrong there during a vacation of mine, BTW). Only now we are going to face a situation where management are trying to cut costs like the way you describe in the place I'm working now, but in the monitoring department... I think it won't take more then half a year before they will decide cutting costs there is not a solution...

Just my .02 (fill in any valuta)

[Alex Ostapenko]

Dear John --

You have received a lot of good responses.

Our environment:
~200 HP-UX servers
~36 Sun Solaris servers
~300 Oracle Databases and other applications

Our support:
6 hands-on UNIX sys admins
5 Oracle DBA's
1 UNIX performance specialist
4 UNIX "design" people/3rd-level support
1 separate Help Desk to field user questions

Things to consider that help streamline support:
1. standard install image and configuration -- if you can ignite a server, have standard post-install scripts, and have standard disk layouts, it makes administering a lot of servers much easier; also, standard set of packages on every server, eg., if you install an updated Perl, make it the same for all servers; if you install Ansi-C, GlanceplusPak, Mirrordisk/UX, make it the same for all servers; if you have a /var/adm/sw mount point for the Software Depot, make it the same for all servers, etc., etc., etc.; also, when ordering new servers, try to get similar model lines

2. LAN console and/or SCM -- this makes remote administration of servers much easier

3. proactive patching and firmware strategy -- pick a strategy (we use conservative, i.e., patches at least 3 months old and quality pack install once per year), but be proactive not reactive; and if possible use the same depot for all servers; having a planned strategy can help avoid unplanned reactive patching

4. proactive security strategy -- have a process for identifying and proactively installing security patches, a process which is outside the annual depot install; have a server-securing strategy such as Bastille, and do it on all servers

5. centralized log and software repository -- designate one server as your "sys admin" server, and create cron jobs to gather your stuff in one place, makes it easier to examine the logs; also have central depot server for distributing patches and other products; having it all in one place (plus with repository backup & a disaster recovery server) will help make other servers consistent and easier to maintain

6. make-net-recovery of every server -- it'll be much easier to reimage if a catastrophic failure or bad patch install should occur

7. control root password access -- so you know who is making changes and what they are

8. document your processes and procedures for doing various tasks; yes, every sys admin should have knowledge of how to do a particular task, and in UNIX there are always at least a half-dozen ways to do it; but, when there's a disaster and the heat is on, it's much easier to check off items on a list (say to replace a failed mirror and restore a filesystem) than it is to try to figure stuff out from memory

9. ... and there are a number of other ideas like this that others on the forum can add

Happy system administrating!

=:-) Alex

[Hazem Mahmoud_3]

In our company, we have 2 HP-UX systems, about a dozen linux systems, and another dozen windows systems.
I'm the Unix administrator, so I basically take care of the core HP-UX systems. The Linux and Windows systems I also administrate, but the workload is divided between myself and a another networking individual.

-Hazem

[Rick Garland]

How demanding are your customers?
How careful are your customers?
How well written are the apps?

These questions (and more) can provide a systems:admin ratio.