2

It's 2017 and I'm looking for some "best practices" guidance with IIS Web Farms across multiple VM servers. Pre-Server 2016 recommendations like

  • Distributed File System-Replication (DFS-R)
  • Application Request Routing (ARR) and Network Load Balancing (NLB)

might not make sense anymore given Server 2016, Clustered Shared Volumes (CSV). Azure IAAS platform also adds their own Load Balancer capability.

TechNet teases at SOFS as a "recommended" solution in the "Practical Applications" section, yet only one bullet point. https://technet.microsoft.com/en-us/library/hh831349(v=ws.11).aspx

"The Internet Information Services (IIS) Web server can store configuration and data for Web sites on a scale-out file share."

What's the latest "best practice", given Server 2016, for both on-premise and Azure IaaS (Not PaaS!)? Is SOFS (active-active) up to the task for IIS Web Farms?

Robert4Real
  • 131
  • 4

2 Answers2

0

You won't be able to deploy DFS-R because it can't replicate open files, and third-party locks like one from PeerLock are both expensive and unreliable. TL;DR: Stick with a recommended solution: SOFS.

BaronSamedi1958
  • 12,510
  • 1
  • 20
  • 46
-1

Server 2016, Clustered Shared Volumes (CSV).

Which ins 2016 are NOT meant to be used outside VERY specific use cases, and no, websites inot one of them. This HAS changed with 2019, but even then I would go for independent farms as much as possible. Yes, a clustered shared volume means nearly similar redundnacny....

...up to the moment you want to make an A/B test or a partial test upgrade and realize all your web servers use the same codebase.

Azure IAAS platform also adds their own Load Balancer capability.

This may or may not be relevant depending on what you do - i.e. I would try to do a lot of load balancing by having independent copies in closer locations once I hit a certain size. There is no one site fits all best practice.

TomTom
  • 50,857
  • 7
  • 52
  • 134