First published on MSDN on Feb 15, 2008
Hi,
My name is Ahmed Bisht. I’m a program manager on the Clustering and High Availability group at Microsoft. Along with a couple of areas of ownership on the failover clustering side, I’m responsible for Network Load Balancing (NLB). In an effort to get closer to our NLB customers, my first post is actually a request for your input. I would love to hear about how you use NLB and what your thoughts are on NLB. Your input would be of tremendous assistance to us as we plan our post Windows Server 2008 releases.
To help you get you started, I compiled a list of questions below. Use it as a guideline. Please don’t hold off on sending your input if the list is too long. You don’t have to answer everything. Also, don’t feel obligated to stick to the list. By all means, if you have other feedback points you think are more worthy to bring up, please do so.
You can post comments directly to this post by signing-in and clicking on the
Comments
link under the title, or send us an email directly through the
Email
through the link in the right pane.
Questions about setup and configuration:
1. How do you setup NLB? With NLB Manager? Locally on one of the cluster nodes? Or remotely? Or, do you use automated deployment tools (like unattended installation)?
2. How many NLB clusters do you have in your environment?
3. What is the range of the number of nodes per NLB cluster? What’s the most common cluster size?
4. Do you have nodes that are part of multiple NLB clusters?
5. What workloads are you load balancing? For example, IIS, ISA, others.
6. Do you commonly load balance one workload per NLB cluster? Or, multiple workloads on the one NLB cluster?
7. What hardware are your nodes connected to? Hub? Switch? What model?
8. Do you have redundant hardware in your environment?
9. What operation mode (unicast, multicast, or IGMP multicast) do you use?
10. Do you run on Windows Server Core (in Windows Server 2008)?
11. How many virtual IP (VIP) addresses do you use? How many are IPv4? IPv6 (in Windows Server 2008)?
12. How many dedicated IP (DIP) addresses per node do you use (in Windows Server 2008)? How many are IPv4? IPv6?
13. How many port rules do you use? Do you use the default port rule (all ports, both protocols)? Or do you create your own rules? Details on port rules: affinity type (no affinity, client affinity, network/class C affinity), equal load weights or disproportionate weights, filtering mode, etc.
14. Are your NLB cluster nodes domain-joined?
15. Does your environment use IPSec?
16. Do you use NIC teaming?
17. Do you use virtualization? Virtual Server? Hyper-V (in Windows Server 2008)?
Questions about management:
18. How do you manage NLB? Remotely? Locally? Using RSAT Client (admin pack)?
19. How often do you add/remove nodes from your clusters? What condition(s) result in those actions?
20. How do you perform maintenance? For example, applying security patches, or service packs.
21. When upgrading from OS version to the next, do you use rolling upgrades to minimize downtime?
22. Do you use MOM (and the NLB management pack) to monitor NLB clusters? Or, do you leverage/build other tools?
23. How do you troubleshoot NLB clusters? Do you find the event log entries sufficient? What other tools do you use?
24. Do you use the NLB WMI provider for customized monitoring/management of your NLB clusters?
25. Do you use nlb.exe/wlbs.exe? What for?
General questions:
26. What are the benefits you see in using NLB?
27. What are the problems, pain points, or areas for improvement?
28. What changes would you like to see in NLB?
I plan to follow up with other posts. So, please let me know if you have specifics topics of interest that you’d like to hear about.
Thanks for your time getting through the list above and thanks in advance for your feedback!
Regards,
Ahmed Bisht
Program Manager
Clustering and High Availability – Microsoft
