3750 stack & scalability

Problem Details: I have an existing  stack of 4 x 3750G switches, which connects servers,
workstation and phones for a client. The company will drastrically grow in the next 3
months from the current capacity to 70 users requiring about 4 network points each. In
addition their server environment will grow, currently using 18 physical servers (2 NIC
each server) and a project to move to virtualisation using HP Blade Servers, connecting at
1GB or 10GE ethernet.

I would like to know what is the best path to make the current switching environment
scalable, maximising the performance and network capacity without necessarily going down
to the Cisco 4500 series switches route.

My idea was to buy a couple of Cisco 3750E-24 series and another stack of 4 x 3750G-48
switches which I will etherchannel to the 3750E. The 3750E will became core switches which
will perform routing, VLANs,  while the 3750G stack will only act as access layer
switch using IP-base images. The existing 3750G stack will be Trunked (etherchanneled) to
the 3750E switches same way as the new stack. By doing this I will have 1 core stack of 2
x 3750E and 2 stacks of 4 x 3750G each.

My plan was to connect, workstations, phones, dealerboards, printers and other devices to
the access layer switches (3750G) while all servers and blades to the 3750E directly
for performace reasons

Is this solution viable and sufficiently scalable?

Please advise

Who is Participating?

[Product update] Infrastructure Analysis Tool is now available with Business Accounts.Learn More

I wear a lot of hats...

"The solutions and answers provided on Experts Exchange have been extremely helpful to me over the last few years. I wear a lot of hats - Developer, Database Administrator, Help Desk, etc., so I know a lot of things but not a lot about one thing. Experts Exchange gives me answers from people who do know a lot about one thing, in a easy to use platform." -Todd S.

Yes I have implemented several similar configurations. All depends upon how resilient the core needs to be, but with two switches you can dual link the access stacks,  servers, etc.
i would add a 3750E switch to the simpler stack and use the 10 gbit interfaces to connect to the core stack of 3750E switches  using cross stack lacp lags for resilliency

3750g                                                          3750g
3750g                                                          3750g
3750g                |----3750E----|                    3750g
MatscoAuthor Commented:
Sidetracked: Looks brilliant, althought the costs would be very high though ? (4 x 3750E ?)

GJHopkins: I was thinking to create 2 x Channel-groups on the cores and create a port trunk to every member of the access layer stack.  I believe the only bottleneck in this scenario would be having Servers on the Access layer switches. All server would have to be sitting on the cores for performance reasons unless I can for Sidetracked suggestion ?
Webinar: Cyber Crime Becomes Big Business

The rising threat of malware-as-a-service is not one to be overlooked. Malware-as-a-service is growing and easily purchased from a full-service cyber-criminal store in a “Virus Depot” fashion. Join us in our upcoming webinar as we discuss how to best defend against these attacks!

Yes, depends upon the bandwidth you think you may need. Often links to the first and last switches in the stack are enough, to lose connection from access to the core you will then need to lose either two links or two switches, and if you lose two switches then you have lost most of the access devices anyway.

If you have a lot of traffic from access to core then the 10Gbit option using the 3750Es can be considered but only you can tell if the required performance justifies the additional expenditure. Remember that the 10Gb slots on the 3750E can support 2 SFP modules which gives you an upgrade path to 10Gbit later if you don't need it all today.

MatscoAuthor Commented:
Thanks, that's clear. is there any issues with etherchannel or spanning tree I need to be aware of with these configs?
Only that if you want to use an Etherchannel negotion protocol PAgP or LACP and the interfaces are on different switches in the same stack then you must use LACP as PAgP doesn't support this.

See article here for explanation


If you have a single etherchannel from the core to each of the access switches than spanning tree isn't an issue as they will both be forwarding.

If you want to run multiple etherchannels fromt he core to the access switch then one will become the alternative path ( assuming the core is the root switch). You can load-share by setting the spanning tree costs differently for say odd and even numbered VLANs but I'd just stick to one etherchannel with multiple links

Experts Exchange Solution brought to you by

Your issues matter to us.

Facing a tech roadblock? Get the help and guidance you need from experienced professionals who care. Ask your question anytime, anywhere, with no hassle.

Start your 7-day free trial
MatscoAuthor Commented:
MatscoAuthor Commented:
It's more than this solution.Get answers and train to solve all your tech problems - anytime, anywhere.Try it for free Edge Out The Competitionfor your dream job with proven skills and certifications.Get started today Stand Outas the employee with proven skills.Start learning today for free Move Your Career Forwardwith certification training in the latest technologies.Start your trial today
Networking Hardware-Other

From novice to tech pro — start learning today.