ThingWorx High Availablity Architecture

[VV_10743662]

Hi All,

My requirement is for TWX HA architecture. I checked the below link from PTC support.

https://support.ptc.com/help/thingworx/platform/r9/en/index.html#page/ThingWorx/Help/ThingWorxHighAvailability/OverviewOfThingworxHighAvailability.html#

I have a confusion that whether Apache zookeeper and Apache Ignite will be installed on each server node along with Thingworx Foundation application or These two will be present on a separate server from the Thingworx Foundation nodes and maintain the HA architecture.

Thanks for the help.

[PaiChung]

Although Ignite and Zookeeper are very lightweight, generally they are installed on their own servers, Each servers a specific purpose and works together as a triad or greater, so you don't want them to be affected by the other if possible.

[VV_10743662]

@PaiChung Thanks for your answer.

Just for Clarification so in the you are suggesting is I should have 1 node which will have Zookeeper and Apache Ignite on 1 server, and 2 nodes for keeping TWX Foundation. So that my application would be always running is in HA.

Is my understanding correct, Kindly confirm.

[PaiChung]

Actually at minimum you would have 2 if not 3 TWX servers. and then 3 or more ignite servers and 3 or more zookeeper servers.

[VladimirRosu]

Whenever you speak about HA, you basically need to define, on your end, how resilient you want to have the setup.

In short, if each component is isolated in its own VM/instance, the degree of failure is higher than if you have the components bundled in less VMs.

The most compact HA ("I want to have the cheapest HA") I believe can be achieved with 3 VMs (if we don't speak about Kubernetes), but this involves a lot of components potentially impacting each other in case of failures. In any way, this would be better than a single node, so it's still good - though not how the theory asks. 2 TW nodes are enough for this type of setup.

And remember that the higher the number of TW nodes, the higher the cluster performance, so you should take performance requirements into account as well.

[VV_10743662]

I checked the below link for sizing for HA components there I found A three-node set, each with 2 vCPUs and 2 GiB RAM would be recommended. But its OS requirement is missingy.

https://support.ptc.com/help/thingworx/platform/r9/en/index.html#page/ThingWorx/Help/Installation/Sizing/thingworx_cluster_sizing_considerations.html#

Can somebody help me defining the Sizing and OS requirement for Apache Ignite and OS requirement for Apache Zookeeper.

Thanks in advance:)

[VladimirRosu]

I believe it's very important to put in the right context the sizing you specified: that 2 vCPU and 2 GB RAM is for Zookeeper alone.

If you plan to put multiple elements in the same VM, as explained in my other post, that VM needs to obviously have the performance requirements of the most higher computationally expensive element (in our case, a ThingWorx Foundation node).

Before going into more details, what is the HA architecture you decided to follow?

In the same link you shared, it's also specified "Review the Apache Zookeeper System Requirements for more details: https://zookeeper.apache.org/doc/r3.1.2/zookeeperAdmin.html#sc_systemReq", and if you look there you'll get the OS requirements.

If you're looking for sizing for Ignite, again, in the same page you'll see test results and specifications for the Ignite node VMs that we used in that test.