Stueyb

Hi everyone.

Im after a little help here. I have seen the power of FreeBSD/Unix and want to bring it into the company. It has come at a good time because we recently audited and have too few licenses.

So what I want to do is merge 3 servers (2 file and print on NT4 and 1 ftp server) onto one machine. Currently these machines are creaking at ~400 - 600 MHz with the exception of one server that is duel Xeon CPU with 1 GB RAM per CPU and ~300GB RAID 5.

Now I could put all this onto the one server and it would easily handle the load, even without kernel optimisations. Only problem is management are scared

So compelling arguments i have put forward are:

1. Its going to save about £2,000 in licencing
2. Less vulnerable to viruses/exploits.
3. More work, less cost. (as in maintainance cycle)
4. Better uptime

Anyone have any other convincing arguments (bare in mind we like pre compiled binaries and the free as in do whatever you want with it doesn't really apply here as we really have no "custom" needs.)

stevencotton

The main problem is having staff that can look after the machine. It may be completely free and super reliable but if you do get a problem and you can't fix it, that's a lot of downtime and that could potentially stop people from working depending on what you use it for. That's the kind of problem management will want to make sure doesn't happen. If you don't have any UNIX gurus there I would probably avoid FreeBSD and choose a supported Red Hat with a good SLA, which of course you will have to pay for.

thundertiger

Stuey, wheres those pics???

BB

DrEvil

Building on what Steven has said, even the support provide by the likes of Redhat is not great, when compared to the likes of Sun, IBM, HP, and even Micro$oft.

We have evaluated the likes of Redhat, and all I can say is we had to lower our expectations on the timely resolution of issues. We have decided to only use Linux for systems we can tolerate downtime on.

For example, a farm of webservers all performing the same task, if one has an issue, we can live without it for sometime as the others take over it's workload.

They may call back in a short time based on SLA, but that doesn't mean they will fix it.

andys

Isn't Solaris 10 meant to be open source

stevem2k

No . Open Solaris will be though.

DrEvil

Apparently - but I heard there are legal issues surrounding this currently due to some of the subsystem components not actually being 'owned' by Sun, just used under licence by Sun. So, the Open Source release is delayed.

But, Solaris 10 x86 would be a better option - although part of the support agreement is that the admin(s) looking after the system are qualified to do - Sun have been known to challenge companies if too many support calls are logged for minor Unix queries... As have IBM and other Unix vendors.

It is important with Unix or Linux, that you get some training, you can wing it, but this can result in some nasty surprises or just poorly implemented solutions that may be flakey or not perform.

DrEvil

So is that their marketing phrase for December - they keep changing their tune....

DrEvil

StueyB - this may or may not be of interest, you can download X86

http://www.sun.com/smi/Press/sunflas...0041115.2.html

Sun plans to make Solaris 10 available for SPARC(R), x86, AMD64 and EM64T systems as a free download by January 31, 2005. Unlike other free distributions which provide only "early access" or "evaluation" free, Sun will provide its commercial grade Solaris 10 product free for deployment. A range of services are available for a fee. Solaris 10 is a vendor-neutral operating system that is supported on more than 270 different hardware platforms from vendors as diverse as Dell, Fujitsu, Fujitsu Siemens Computers, IBM and HP. Sun also unveiled a new advertising campaign and game-changing business model in support of the Solaris 10 launch.

stevencotton

Our SLA is actually with Dell, they've been quite good so far, of course the only problems we're likely to require assistance with are very low-level hardware incompatibilities. Everyone here is a full guru.

I do like Sun kit as you know and you get fantastic support, but the hardware is very expensive which may negate any cost savings, all OEMs will be the same.

orbv

Have to agree with Dr Evil.

You may get likely and never run into any bug OS issues but if your unlikely can you debug and fix the code in a good timeframe? There is a big difference between writing application code and investigating crash dump then fixing the kernel code without effecting anything else and corruption your data thats not backup. Unless the box is redundant or unimportant you will need someone to back the support onto.

The price of the software licenses is small compared to the support costs so saving yourself £200 is not worth it. Add to this the cost of retraining multiple people and your out of pocket in a big way.

At some point Solaris will become open source. Around a year ago Sun brought a license from SCO to allow them to make more of the code available but some of the code will not be released so its unlikely you will be able to compile a full Solaris kernel.

Does it matter if you dont have the code? 99% of people would not know what to do with it. So Solaris will join some linux or bsd distributions and become free but this is only saving you max 200 quid. Its all just marketing talk.

Anyway, in general if you can download it for free it wont have any worth while support.

DrEvil

*bow*

Full Unix geeks - surely! Just like my office

(excluding me of course)

stevencotton

No, some have cracked physics and really do float a foot off the floor like the guys in Dark City

DrEvil

Medication quite strong then! or should that be meditation! groan..

stevencotton

Don't start another Amiga thread I'm going to bed

DrEvil

PS. sorry Stuey

Lum

There's plenty of 3rd party companies and contractors out there that sell support contracts for linux installations, it may be worth investigating that route.