Half a hero biscuit for Microsoft
Enjoyed installing OCS? Good, you may be reinstalling soon
SaaS data loss: The problem you didn’t know you had
Sysadmin blog Neat as it is, Office Communications Server (OCS) is complicated. It has grown beyond being one application server into a collection of interlinked but separate application servers, each with their own requirements.
There is a front-end server component, Web Conferencing, A/V Conferencing, Application Sharing, Communicator Web Access, and a third-party application add-on server that runs components like Microsoft’s Response Group Service. Each of these is their own entity with their own requirements. Each is capable of working or breaking entirely independently. The latest version of OCS also predates the current generation of Microsoft server products so, if you install things out of order, it all blows up.
In its incarnation as an instant messaging extension to Exchange it was a tickbox install with not much thought required. Live Communications Server 2003 was more complicated, but still something that I could install over lunch. This contrasted with the mental suck that resulted from Live Communications Server 2005 (LCS2K5).
LCS2K5 was one of the most miserable products I have ever had the misfortune to administer.
I have been through the fire thrice: by the third time I had caught on that putting this thing inside a virtual machine was a good idea. That way, when it blew up, I could restore a working copy. A chunk of the blame goes to Windows Server 2003’s numbingly awful Certificate Services. Windows does not do certificates in a straightforward manner. Even putting the certificate problems aside, LCS had a lot of components. There was so much of it that “where to begin” was overwhelming.
Along came OCS, with a little wizard that abstracts all of the pain of Windows Certificate Services from you. Push button, receive bacon: I was impressed. The person at Microsoft who realised how miserable it was to fuss with Certificate Services - just to install a minor piece of server software - deserves a hero biscuit. It’s an innovation which Exchange 2010 and IIS 7 thankfully share.
OCS is still an enormous suite. With that in mind, you would think that the OCS installer was modular. But while you can break components of OCS independently from one another, be prepared to reinstall the entire thing when you do. Similarly, if during the initial setup you install only some components, be prepared to live with your decision. Try to install others at a later date and the entire thing tends to die, requiring a complete reinstallation before it will cooperate with anyone.
My travails with OCS are similar to working with LCS2K5. We get licences with our Action Pack, so my CTO makes me use it, but it’s too big and too complicated for what I want it to do. It takes a lot of setup, seems to need almost as much maintenance as Exchange, and is exceptionally fragile.
I can see the value of OCS: even with the basic presence integration into Microsoft Office, the application sharing and the basic SharePoint integration that we use it's miles ahead of a simple Jabber server or Skype client. But it fits a classic Microsoft pattern: fantastic if you are large enough to devote an administrator to it. Like so many products that Microsoft makes, there isn’t a “dumbed down” version available for the SME.
Thus I am torn. For a really small shop looking for corporate instant messaging, the advice is easy: don’t bother with OCS. Slap a jabber server together in 15 minutes and go have a cup of tea. For a large enterprise, OCS competes well against IBM and Cisco, and you’d be foolish not to consider it. The mid-market - folks like us who are too big to use Microsoft Small Business Server but too small to devote people to specific application servers - are left out in the cold. Again. ®
COMMENTS
@darren 16
Nowhere did I say it wasn't an excellent application. I said it was a pain to install and maintain. As for your impression of sysadmins, I can't help you there. I am sorry to report that we are human beings and as such we tend to prefer applications and products that make our lives easier. I don't think we are any different from anyone else in that regard. I should also point out that despite OCS bieng a pig to work with, I did deploy it. I did so because it would benefit my users more than alternatives that were easier on me.
None of that changes my desire for it to be easier to administer.
Clouds
He may be paranoid...but I am not yet convinced he's wrong. You have to go a long way before you convince me that putting your data in someone else's care is a good idea. It boils down to trust, and I have very little of it for soulless Megacorporate entities.
Entirely apart from the data governance issues, hosted services leave you with noone to flog when it goes boom. SLAs are worthless, and don't actually cover the costs of downtime to your business. What's more, you have no control over if/when your hosted provider will pull a Zune on you and withdraw or radically change the service. What is someone in my situation going to do? Sue Microsoft? Please. The legal fees alone would be ruinous.
No, cloud anything is an absolute last resort for any business critical function, unless you are dealing with a provider that lives and dies by their service levels. A corporation like Microsoft has no incentive to do anything other than tell you how it's going to be, and expect you to like it.
Even if you could argue that Microsoft had some minor incentive to do well by it’s customers, it has proven time and again that it doesn’t care about any organisation below a certain size. Do you honestly thing this suddenly changes because it’s a hosted service? Balderdash. Unless the tco over an expected application lifetime of 4 years was SIGNIFICANTLY lower, the risks outweigh the rewards. Oh, and “but the keep you on the latest version” isn’t a pro, either. I want to deploy a server fleet and have it function with no major changes – or user retraining – required for at least 4 years at a time. I don’t upgrade every iteration of anything unless there is a DAMN good reason.
If it isn’t broke, don’t fix it. “Because it’s newer” is never a good enough reason for ANYTHING. So what the benefit of hosted services? Where are those advantages or TCO price delays over a four year application life that make it worth it?
At 75 users, I just don’t see it.
dz-015
Point taken. I do use Redhat a lot, and VMWare…a few others. I will make a point to yak about them some more. I am sorry that it’s been so Microsoft all the time recently…but we just replaced our whole Microsoft domain.
4 cities, 5 sites, 5 DCs, Exchange server, WSUS, OCS, a half dozen SQL servers, and the list goes on. Two sysadmins and a bench tech. 250 VMs and about 100 physical machines. 2.5 days; with as little interruption of service as possible, and no…we didn’t have enough hardware to run both networks in parallel for the changeover.
All of it had to be done without disrupting the Linux side of the network for more than a few minutes at a time. Total sysadmin uptime: over 100 hours per wetware unit over the course of the 3 days of changeover and two days of post-doomsday tech support.
Oh, and they are letting me blog about it!
IT infrastructure monitoring strategies
Agentless Backup is Not a Myth
Top 10 SIEM implementer’s checklist
Steps to Take Before Choosing a Business Continuity Partner
Enabling efficient data center monitoring
