Recommendations for server redundancy etc?

[swpmre]

Hi all,

 

I work for the city's Adult Ed service - multiple sites scattered over a large area.

 

We've had multiple problems in the last few weeks with power failures and other problems at the site which hosts our main servers.

 

We have a back up Domain Controller at a different site, but currently do not have any redundancy for our file / print server. When we lose connectivity to our main site for a few hours all our staff and students lose access to their files etc.

 

What do you recommend as a solution? Ideally I'd like some sort of server replication so that if we lose one site, users (relatively) seamlessly can access their files at a backup / alternate server.

 

After the problems of the last week, we may find there are funds to do this, but I am looking for lower cost solutions.

 

Thanks,

 

Martin

[john]

DFS sounds like it could be what your after but you should look and plan into it. You would have file servers on each / a number of sites and use DFS so it automatically replicates the files in the DFS shares to other sites and allows you to connect to the nearest to the site you are on

[PotNoodleTech]

There are lots of things you can do to mititage the main site going down, some fairly cheap (DFS) some more expensive (Vmware with High Avilability).

 

I would however suggest you try and address the main issue - why is the main site with all your servers having so many problems with power? Do you not have adequate UPS (unninteruptable power supply) or generator backup? How often and for how long are the backouts?

 

If it is a case of investing in UPS to give you a 1 hour uptime "buffer" during a power cut, then that is fairly cheap and easy to impliment. If it is a case of having UPS and and generator on standby then this is much more complex and expensive. I would ask - have there always have been severe power problems, and will there continue to be? If yes, Maybe that "main site" is not really suitable for being your "main site".

[swpmre]

All these questions are ones we (and senior management) are now addressing. The problem is that no one can tell us why the power is going down. They're building a new tram line outside, and that's the logical reason, but naturally they deny its their fault. This work is scheduled for another 18 months, so it may be a need to move sites is crucial.

 

We have UPS, but the power cuts are taking out all the local infrastrucure, so comms, telephones etc all die and we loose all connectivity. Generators are unlikely due to cost. It is perhaps easier to consider alternate sites for our servers anD DFS solutions as suggested earlier.

[SkreeM1980]

As with all things IT there are many options available to you.

 

DFS may well be a quick win, it depends on how your sites are spread out/setup.

 

You don't mention what Comms you have between sites, but when i see lots of small sites what springs to mind is looking into cloud/hosted/COLO solutions. If you are already using the internet connections at your sites to create VPN's to the site housing the servers why not move that equipment to a datacentre where they have all the power/connection resilliencey in place, and your just having a little space in a rack for your kit. Any solution like this is of course budget dependant, but it could work out better than a siginificant capital outlay to put the redundancies in to pay the fees.

 

No solution is fit all ... look at the options and decide what works best for you.

 

Any resilliency is allways a balance of budget vs availability, and how many of the 'What If' situations you want to cover.

 

Good Luck

 

Skr

[Millgate]

Well firstly I’d suggest looking at a colocation facility to get around the power issues, a hosting facility should have multiple redundancies in place, this would be the first place to start looking at potential improvements.

 

An alternative may be to install small (cheap) local Domain controllers on each site to provide local print, active directory authentication and file shares, these could be staging points for DFS system to transfer data to the “host servers”.

 

This would work in one of two ways, the host servers could replicate data down to the local servers, and in the event of an outage local users use the local data stores for replicated data, or the local sites use the local servers for the main bulk of work which is then replicated back to the hosted platform.

 

Active directory authentication could be distributed to the local DC’s allowing local logons if the datacentre fails.

 

Two initial lines to progress here, its more than likely the datacentre would be the best and easiest place to start at as the infrastructure is already in place.

 

Luke

[swpmre]

Thanks everyone.

 

As someone said, this is a question of balance. And not just of cost. Some of our sites are very small, with a tiny number of users and not used for some days of the week. We might have to balance out occasional lack of network use compared to cost of installing local DCs etc. Either way, the DFS solution seems to be on the right scale, though printer availability is still an issue - we currently have a single central print server. That might need to be reconsidered.