SRX

We have serveral SRX240 clusters at various sites, mixture of high memory sites SRX240H sites and low memory sites SRX240B sites and the question I have is in the event of hardware failure we have high memory hardware spares but could we in event of failure introduce a SRX240H into a SRX240B cluster?  We are not using any of the licensed features that the high memory platforms are capable of, the clusters are jsut used for basic firewalling and VPN's.

 

Many thanks in advance

Hi,

 

It should work but you have a number of things to bear in mind:

 

• The B model originally had 512MB of RAM and the original H has 1GB, I am not sure if the CPUs are identical speed.
• In the event that you end up using more than 512MB of RAM (say a failover to the H node happens and is left in place), how will the B node behave?  Strangely, I would assume!

I'm sure the answer from Juniper is that they do not support it.  It should "work" though as in the Cluster will form, to be honest, I have never tried it.

Thanks for the reply, much appreciated.

 

As we would only be introducing a a SRX240H node into a B cluster it should only have to pick up the load from a B memory cluster, so we would be always adding a node with more memory available than the other, my only concern is like you say the cluster falling over to this node and then falling back to a B memory node and the B node not liking this.  

 

I guess we will have the same kind of issues now with the new juniper SRX240H2 and SRX240B2 range as the memory on both the version 2 models is more than version 1 so if the memory mismatch is an issue in a cluster and not supported we will have to look at having lots more hot swap spares available.

 

Thanks again for your help on this.

In my humble opinion cluster will not form between two different model firewalls, as basic requirement for chassis cluster is identical hardware model

 

Reference link

 

http://kb.juniper.net/InfoCenter/index?page=content&id=KB15504

 

 

 

 

Please mark this as accepted solution if it works for you

A kudos is a good way of appreciation

 

Kashif Nawaz

JNCIP-Sec ,JNCIP-Ent

JNCIS-Ent, JNCIS-Sec

JNCIA-Junos

 

 

Dear 

 

I think it will work , as the hardware is almost the same , only difference is the DRAM (1GB for BM and 2GB for HM ), so , this will affect your concurrent sessions only (128K for BM and 256 for HM) , considering you will not use any licensed feature like UTM/IDP that requires HM model, so if you make sure you will never reach the 128K threshold , so I don't think you will face an issue with the session creation and synch between the cluster nodes, this only theoretical suggestion and cannot be considered as confirmation, you can test it in lab before implimentation in live network.

 

Regards

 

Many Thanks for everyone's input on this, I guess the answer is as Red1 mentions to test it out in a lab and see if it works, unfortunately I only have access to High memory models in our test lab, the only base memory hardware we have is in production so If anyone has the ability to try this out in a lab I would be interested in the results.

 

Thanks again for all the replies

 

 

This definitely does work, we had an H2 recently fail (no power) and ran with a SRX240H as the replacement node1 in an active active cluster for about 3 months without issue

Hi all,

 

As Route Champ above already correctly mentioned, both chasss cluster nodes are required to be of the same model. In a cluster it is not supported to mix between B/H/HE models.

 

Even if a lab test would show that it may work now, it is not an officially supported configuration and may not work anymore later on in a higher Junos release.

 

Thanks,

Casper