In the document regarding replacing a ESA in a cluster it states that it needs to have the same asyncos build.
We have two old ESA that we are going to replace with new ones. The old ones cant be upgraded to the async version that the new one comes with. Do we have to build a new cluster. Or should the line "same asyncos build" read "at least the same build or newer than the existing esa:s?"
The question I have for you is, what model is the old hardware? If this is x60, and it only goes to 8.5.7, you'll need to download and deploy a virtual ESA (on the 8.5 family) and join that to the old HW cluster. Then you'll have the ability to continue the vESA running the configuration, upgrade AsyncOS on the vESA, and have it ready to match AsyncOS version that is running on the new HW that you are going to replace the old with.
Configuration has to be the same AsyncOS revision. If not, you'll be facing hand configure/copy & paste of the configuration - which is not always easy (or recommended). (*Unless you want to get a fresh start on the configuration)
Running a vESA can be done on an ESXi host, or even using Fusion (OS X) or VMware on PC. If your license is current on the old HW, you can get the needed .xml license file.
See this for more info on getting started with the vESA:
No, for the C370 you are fine to upgrade past the 8.5.7 family. It is an End-of-Life build, so it would be recommended to upgrade it anyhow - in order to continue getting updates for all services. The x70 series HW can upgrade all the way through 11.0.x. x70 final version that they are limited to is the 11.0.x, so they will not see 11.1.x.
That will make it easier in terms of upgrading, you'll just need to make sure that the x70 and new HW (x90?) will be running 1:1 AsyncOS releases when you are ready to implement the replacement. Then it is just adding x90 to the x70's cluster...