Logo 
Search:

MOSS Forum

Ask Question   UnAnswered
Home » Forum » MOSS       RSS Feeds

High Availabily for MOSS: MS NLB or dedicated hardware?

  Asked By: Fallon    Date: Feb 27    Category: MOSS    Views: 2212

What is the preferred method of load-balancing MOSS 2007 Web Front
Ends? Is it Microsoft Network Load Balancing or dedicated Cisco
hardware? What are the pain points for MS NLB?

Share: 

 

7 Answers Found

 
Answer #1    Answered By: Chelsey Watts     Answered On: Feb 27

The main one i know of is amount of users and hard cash

If you can afford it use a dedicated load balancing device for
throughput etc - if you have limited users ms nlb  is fine (i.e. 5,000
users use hardware, 100 users use software)

Other than that, ms nlb is fine - i've not really heard of problems with
it in smaller implementations

 
Answer #2    Answered By: Nagesh Maulik     Answered On: Feb 27

I have heard that MS NLB is the way to go. I have heard this from Microsoft
themselves (obviously) and from other non-MS folks out there. I do know of an
organization that uses a hardware  solution and it's been proven as 1) costly 2)
very difficult to get setup and 3) overkill for the purposes of MOSS WFE.

I currently have 1 WFE in my Prod environment but I will employ the MS NLB
methodology as soon as I'm ready to tackle that.

 
Answer #3    Answered By: Irving Hurley     Answered On: Feb 27

i agree, but if you have a whole bunch of concurrent users it may
be cheaper to put the load balancing onto a separate dedicated device
rather than adding more web front ends to handle the load

For most implementations ms nlb  is fine (is what i meant to say) but for
the larger ones hardware  can be good - especially if you have multiple
solutions (other than moss) that you want to load balance also

I guess my real question should have been .... how many users doing how
much work ?

 
Answer #4    Answered By: Asia Meyers     Answered On: Feb 27

We also use our hardware  load balancer for other routing jobs, such as
routing users to an outage page when a service is down, and for some URL
rewriting tasks. You may want to consider how hardware and software load
balancing approaches compare WRT more than just load balancing.

 
Answer #5    Answered By: Shelley Reese     Answered On: Feb 27

F5 and Microsoft have collaborated on a highly effective solution for
intelligently directing traffic to SharePoint Portal Server 2003 & MOSS
2007 deployments with the BIG-IP system from F5. Microsoft SharePoint
Portal Server 2003 enables enterprises to develop an intelligent portal
that seamlessly connects users, teams, and knowledge so that people can
take advantage of relevant information across business processes to help
them work more efficiently.

Organizations using the BIG-IP system benefit from mission-critical
availability, intelligent traffic management, simple scalability, and
enhanced security for Microsoft SharePoint Portal Server 2003
deployments.

Information on the Microsoft SharePoint Portal Server 2003 -
www.microsoft.com/.../default.mspx
<www.microsoft.com/.../default.mspx>

For more information on the BIG-IP system, -
http://www.f5.com/f5products/bigip <http://www.f5.com/f5products/bigip>

 
Answer #6    Answered By: Vinay Thakur     Answered On: Feb 27

I'm currently running NLB for our corporate intranet (4200 users)
and for one of our manufacturing divisions--both WSSV2 & SPS2003.
It is an adequate solution.

The pain points I've had are

1. Monitoring--if IIS goes down, NLB doesn't automatically switch
to the other node. It takes a node "out" when the TCP/IP heartbeat
is broken. BigIP has sophisticated monitoring that will
2. Network Broadcasting--NLB does some broadcasting on its local
segment, which got our network guys really upset. They put our NLB
pair on its own switch.
3. load-balancing  method--BigIP allows you to control the kind of
load-balancing being used, such as "round robin." After hours of
analysis, my coworker and our network folks were unable to determine
the exact method  being used by NLB--he described it as the "wierd"
method.

I am looking forward to putting our SharePoint sites onto our F5
BigIPs in the near future.

 
Answer #7    Answered By: Christen Roberson     Answered On: Feb 27

Here is my 2 cents since your planning to implement F5 in your
environment:
The Key components of our architecture:-
- Two load-balanced front-end web servers running the Web / Search
components
- Dedicated Index server
- Two servers running SQL Server in an active/passive cluster mode
- Hardware based network load balancing using F5 network's BIG IP
device
- F5 web accelerator module for enhanced performance

Above design displays the redundancy at various levels as follows:
- Web server –> active/active (Load Balanced and Fail over)
- SQL server –> active/passive (Fail over)
- Big IP –> ( active/passive) Fail over

Reason for us to select F5 as our NLB solution:
The use of F5 Traffic redirector as load balancing and failover
device, the user requests made for portal server will be redirected
to F5. Local redirector will take care of distributing the load.

F5 Web Accelerator is used as an integrated software module on BIG-IP
Local Traffic Manager platform. F5 WebAccelerator is an advanced web
application delivery solution that provides a series of intelligent
technologies designed to overcome problems with browsers, web
application platforms and WAN latency issues which impact user
performance.

Above all we are using HP Openview for monitoring of all the
SharePoint services as mentioned my HP OVO SharePoint smart Plug-in.
In the event of any services, going down in terms of IIS, cluster
servcies, SharePoint timer services etc, we have the monitoring
device put in place to wake-up everybody.

 
Didn't find what you were looking for? Find more on High Availabily for MOSS: MS NLB or dedicated hardware? Or get search suggestion and latest updates.




Tagged: