One of the better foundational presentations on what is involved in administering SharePoint (the good and the bad) was conducted by SharePoint MVPs Shane Young and Todd Klindt at the SharePoint Conference this week in Anaheim, CA. While I've seen Shane and Todd present this session (or a variation) several times, their presentations are always worth seeing again. They do an incredible job of keeping people entertained while sharing their extensive experience in supporting, migrating/upgrading, and configuring a SharePoint server.
 

Shane (@shanescows) and Todd (@toddklindt) have experienced just about every pain and made every mistake (mostly the fault of Todd, Shane points out) that you can go through on the SharePoint platform, and have first-hand knowledge of just about every possible configuration. They know what they're talking about, and they deliver the material in a very entertaining way, balancing their content for the savvy admin as well as the new admin trying to understand what it all means.

At the center of this presentation is a walk-through of the deployment process, step-by-step, sharing best practices along the way. They explained the three tiers of SharePoint topology: web-front ends (WFE), or the server that hosts the IIS sites; application servers, or the servers that run all of the various SharePoint services; and database servers (including a quick discussion on Microsoft's forthcoming version, Denali, which Microsoft demonstrated in the keynote).

Noteworthy Topics Covered

It's difficult to put the barrage of content they covered into a neatly packaged narrative, so instead I'll outline some of the topics covered and best practices instilled:

  • Discussed SQL aliases, which Todd has blogged about here.
  • Covered various topologies, with an overview and discussion around each of the different architectures, with advantages and disadvantages of each. For example, there was some great dialog around "carry-over" topologies from 2007 environments, and how they work with SharePoint 2010, such as the 4-server "Quatro" design, which was considered the "medium"' server topology in the MOSS2007 world, but requires SP2010, WFE and App servers to be installed on three of the four boxes, with the 4th as SQL server to get the best performance and redundancy on this configuration. Their message -- while you can do it, it’s not recommended.
  • Discussed variations on expanding the farm, sharing their "cheat sheet" on the number of additional servers needed to add other service applications, such as a BI group running Excel Services and Performance Point Services, or a search group running index and query. There are too many configurations to count, so it’s best to do your research and understand best practices for each service app being added.
  • Don't skimp on RAM. If you go with the minimums, the server and SQL need to spend more time flushing out cache. Go big, as it will directly impact (improve) end-user performance.
  • Shared their "Truth on Virtualization" for production systems:
    • Virtualizing web-front ends is fairly easy, not a bad idea, and works for most designs.
    • Virtualization of the app server is mostly OK, although search is impacted.
    • Do not virtualize SQL. All performance comes from SQL. They shared some horror story of problems with one VM taking down an entire host due to space limitations. Todd used the great line "Don't thin provision SQL," as it is the most critical aspect of SharePoint. He then amended his statement, saying you can virtualize SQL, but most people poorly virtualize SQL. Know what you're doing (at which point he shared the TechNet hardware and software requirements).
    • If you set it up correctly, with everything but SQL virtualized, SharePoint doesn’t care if SQL is not virtualized. It doesn't know, doesn't care.
  • For development environments, go ahead and develop on any hardware you want, but stick with the recommended minimum of 8Gb RAM, but you can also install on Windows 7 or Vista, but must be 64 bit, and NEVER use this for production (MSDN link).
  • They walked through the four stages of a good (new) install:
    • Use the Prerequisite Installer (prerequisiteinstaller.exe). It fails a lot, and if so, just restart, or refresh server and restart. Eventually it'll go.
      • Adds IIS and app server role to windows and configures them.
      • Downloads and installs all necessary software.
    • Install the bits (setup.exe).
      • If you have any hot fixes, service packs, language packs, or office web apps, now is the time to install them.
      • Straightforward, but watch out for standalone server installs, as they are evil.
    • Configure the bits (grey wizard/PowerShell).
    • Farm configuration (white wizard or PowerShell/Central admin).
  • Service packs are always a good idea, as they often release new functionality as well as fixes, but always test first.
  • Hot fixes and cumulative updates should only be applied if they fix a problem. Test, test, test. They are not tested by Microsoft as the service packs are tested, so be careful that they solve specific problems you are experiencing.
  • Oh, and don't forget to turn on outgoing email. Lots of people forget.

For More Information

While there's a lot of great advice here, I'm barely scratching the surface of what Shane and Todd covered in their 75-minute session, and this was only Part 1. Later in the day they presented Part 2, which continued an even deeper dive into SharePoint administration essentials. For more information on SharePoint admin tasks, and for great wisdom from Shane and Todd, you can visit their blogs (Shane, Todd) or check out the deep bench of experts at SharePoint911.com.

Editor's Note: You may also be interested in reading: