When to add another server?

  • Thread starter Thread starter JM
  • Start date Start date
J

JM

What are some standards, benchmarks, or rules for determining when another
server is needed? I realize it's not a clear-cut issue, and an infinite
number of circumstances come into play, but I'm just looking for some
general direction.

In one particular case, this is the hardware:

Dell PowerEdge server, ~2ghz Xeon cpu, 1gb RAM
Win2k Server, SP4
3 Seagate Cheetah HDs, 40gb, RAID level 5
Benchmark DLT Tape drive

The network supports approximately 15 workstations.

This server acts as the DC and runs the primary enterprise app - a medical
billing s/w.
It also runs SQL, Veritas backup, and Symantec security (Client Security
3.0, I think)
Two users connect via TS.

There are no speed problems, and normally the system uses about half the
RAM. The only ongoing issue is the print spooler sometimes gets huge, and I
have to stop/start it. I just became aware of this, and I think it might be
related to the TS settings.

If there were more workstations, I would want a dedicated DC, but in this
case the server's multi-tasking doesn't appear to be hurting the
performance. But, should we be thinking of another server?

thank you,

jm
 
The only reason I would think of adding a second server is for redundancy
and to spread out some of your workload. It sounds like you're fine
though...

--
Thanks,
Brian Desmond
Windows Server MVP - Directory Services

www.briandesmond.com
 
What would happen to this business if you had a hardware failure? Users
would no longer be able to logon except in cached mode, you have no fault
tolerance and if this server fails and can't be resurrected and the hardware
is old you could have yourself some real issues.

You should get a second dc up and running, even if it is on a bit older
hardware.

--
Paul Bergson
MVP - Directory Services
MCT, MCSE, MCSA, Security+, BS CSci
2003, 2000 (Early Achiever), NT

http://www.pbbergs.com

Please no e-mails, any questions should be posted in the NewsGroup
This posting is provided "AS IS" with no warranties, and confers no rights.
 
Paul Bergson said:
What would happen to this business if you had a hardware failure? Users
would no longer be able to logon except in cached mode, you have no fault
tolerance and if this server fails and can't be resurrected and the
hardware is old you could have yourself some real issues.

You should get a second dc up and running, even if it is on a bit older
hardware.

Listen to Paul. Most people 'over-buy' their DC hardware
anyway.

Lose you last DC and you lost the ENTIRE DOMAIN plus
any child domains in that forest.

A $500 (or less) PC can make a perfectly reasonable
second (or first) DC for small domains.

Matter of fact, a throw-away, otherwise obsolete machine
can do quite fine for small domains.
 
Excellent point. And a very timely one, because I just spent the better
part of 2 days (the MUCH better part - all but 4 hours of bad sleep time)
resurrecting a small network in a very similar situation - one DC, failed to
boot, network completely down, with no failover plan. See my other post
"Please help with crashed server."

Please point me to good documentation on best practices for creating fault
tolerance in a small network.

thank you,

jm
 
JM said:
Excellent point. And a very timely one, because I just spent the better
part of 2 days (the MUCH better part - all but 4 hours of bad sleep time)
resurrecting a small network in a very similar situation - one DC, failed
to boot, network completely down, with no failover plan. See my other
post "Please help with crashed server."

Please point me to good documentation on best practices for creating fault
tolerance in a small network.

I am not big on "written best practices documents" but
prefer to pass on design concepts and strategies so that
anyone can create their own solutions that fit their needs
so....

Two of everything is the general rule as a start.

Then another set of everything "somewhere else".

This may sound simplistic but merely having two
DCs with all necessary services like DNS is a great
start.

Then put another DC across the network somewhere
so that if the building burns or floods (see also New
Orleans) or another catastrophe occurs you can
continue.

For everything that cannot be duplicated, make
timely* backups.

*What is timely? Simply ask: Can I afford to
throw away one week, one day, one hour, etc of
my business? Once you reach the level you can
tolerate losing (or ask the actual business people)
then make backups at least more often than the
level at which you cannot tolerate it.

You can play with RAID, dual CPUs, dual power
supplies etc, but many times you are just better off
with anther entire computer.

For $1000 a year you can rent a NICE computer
at GoDaddy.com or some other web service provider
but there is no rule that says you cannot use this for
a spare DC (properly firewalled) or backup location.

For many people a removable USB/Firewire hard
drive may make more sense than trying to use backup
tape.

First make it work. Try to keep it simple and 99.9999%
effective rather than "we'll do it perfect tomorrow etc.
Then improve the working, simple solution whenever you
can.
 
An idea I have not yet implemented with my customers is a Virtual Server on
my premises that runs Replica DC's for each client. Set them up in a
different site with periodic synchronization at least every 7 days. If done
manually, VPN into the main network from the VM, initiate replication using
repadmin (or any number of other tools), ensure it completes, then logoff
and shut down the VM for another week.

In the case of catastrophic failure (ranging from their server dying to
their building burning down or natural disaster) their entire AD is backed
up, online and available with just a few minutes work on your end. Totally
separate from any other backup plan you may have in place. I think that you
might even be 100% within your license rights since this would be a "cold
backup" machine.

Something to think about.

--
Mike Shepperd
Sunfire Solutions LLC
Seattle, WA

**********************************************************
"This posting is provided "AS IS" with no warranties, and confers no
rights."
**********************************************************
 
Well it is recommened that you have minimum of two DC's in every domain.
This ensures access in the event of troubles. I would also encourage you
two have Active Directory integrated DNS (The DC's woudl be providing DNS
services) and make both DC's GC's (As long as this is the only domain in the
forest).

Below is a link for Best practices as well.
http://technet2.microsoft.com/Windo...176a-4592-bcde-a61e733579301033.mspx?mfr=true

--
Paul Bergson
MVP - Directory Services
MCT, MCSE, MCSA, Security+, BS CSci
2003, 2000 (Early Achiever), NT

http://www.pbbergs.com

Please no e-mails, any questions should be posted in the NewsGroup
This posting is provided "AS IS" with no warranties, and confers no rights.
 
Back
Top