HPE SimpliVity

SimpliVity constant bandwidth between OVC's

 
decstewartje
Occasional Advisor

SimpliVity constant bandwidth between OVC's

Hi folks.

I am posting this out of curiosity. I am new to SimpliVity and deployed a 1+1 (small) setup in our envirinment. The 2 nodes are connected via a VPN tunnel, and so far working great. I have noticed that the OVC is very chatty, utilizing a minuimum 4-5MBs constant bandwith between the two. At one point it was as high as a sustained 20MBs.

I am fully expecting some high bandwidth during an offsite (after hours) backup, but this constant bandwidth is used when there are no backups being performed at all at either node. Confirmed that all backups are in a "protected" state, not saving/sending.

What could possibly be happening where the stacks need to be so demanding on the network? Is it safe to throttle port 22122 during office hours?

4 REPLIES 4
tonymcmillan
Frequent Advisor

Re: SimpliVity constant bandwidth between OVC's

Based on what you've described, the system is not designed to work that way. To ensure ESXi HA functionality, a copy of each VM is maintained on both hosts. With SimpliVity, a 2-node configuration requires the nodes be connected by the two 10GbE interfaces. Traffic between OVCs is not just for backup, it's also replicating the VMs as well which can result in large amounts of data being transmitted and in both directions.

To support your remote configuration, you need to have a 10GbE link with no more than 5 ms latency between the nodes.

The supported solution for your implementation is two nodes at your host site and a third node at the remote site, a 2+1 configuration.

I hope this helps.

decstewartje
Occasional Advisor

Re: SimpliVity constant bandwidth between OVC's

It is not configured for HA as it is a 1+1 (and supported). I am aware of the 2+1 configurations. In a 1+1, there is no replication of VM's.

The goal was DR not HA since we can take the downtime as opposed to the additional cost. 

tonymcmillan
Frequent Advisor

Re: SimpliVity constant bandwidth between OVC's

Understood, sorry for the confusion. Given that we could have a similar use-case with a client in the future, I am also interested to hear about the cause of the traffic you are experiencing. 

Chris_Web
Occasional Visitor

Re: SimpliVity constant bandwidth between OVC's

Hi,

We have actually a 48 nodes federation (still growing) running on 4.0.1. All are minimum a 2 nodes cluster and recently we notice that each nodes were sending around 2500 UDP pakets on port 22122 using the L2 IP address of the "Federation network" on the "Management Network".

So even if you don't have HA or replication between the 2 nodes, they will discuss a lot.

That's also why HPe white papers mention the Full Mesh and Hub-and-talk tolpology. 

IT is so important to implement the hub and talk topoligy if you are running nodes on ROBO sites.