HTML5 Gateway Performace

[BrandonP2]

Looking for some general information. We are fairly new running our Parallels farm and the direction we are headed is going to put a lot of users on the HTML5 interface using Chromebooks, Chrome boxes, etc. Is there any best practices for a larger HTML5 deployment? I want to make sure performance is the best it can be. Do we need additional gateway servers? How much of the processing runs there? I'm expecting about 100-150 HTML interface users in the near future.

Also a question on the load balancing...We are using 2 HALB servers and 2 gateway servers. I've noticed that one of the gateway servers is running about 40%-60% CPU and the other is completely idle. Is there something I need to check there? Doesn't seem to be load balancing very well. Is there a way to see how many connections each gateway server is handling?

Sorry for so many questions but really looking forward to finishing up our rollout of RAS. Thanks.

 

[Valentin@Parallels]

Hi @BrandonP2
We do not have any best practices for environments based on HTML5 Gateway connections, however I'll try to provide you with any relevant information that you would like to know.

I'm expecting about 100-150 HTML interface users in the near future.

Could you please clarify if you intend to use HTML5 Gateway connections only, or if you expect up to 150 HTML5 Gateway connections alongside with other users who are using alternative connection methods like Parallels Client or Web Portal? It is important to measure total expected load first.

Also a question on the load balancing...We are using 2 HALB servers and 2 gateway servers

Are there any reasons for having 2 instances of HALB running simultaneously? Basically one HALB should be enough to handle 2 Gateways.
As for CPU load on Gateways I would suggest checking that both Gateways were added to HALB (In RAS Console - Farm - HALB - "Devices" tab).

Will be looking forward to have your reply so we can proceed our discussion regarding HTML5 Gateway connections.

 

[BrandonP2]

We are going to be using the html gateway and native clients. About 250-300 users total. We are running 2 HALBs for redundancy, we are a large healthcare provider and run 24/7 so uptime is very important.

What I noticed with the gateway servers is that for some reason, the ras management console shows server 1 is using 100% cpu for some reason (It's not, I logged in and it's idle) so I'm assuming it's not going to send any connections it's way.

 

[Valentin@Parallels]

Hi @BrandonP2
Thanks for environmental description.
As for environment setup we usually recommend having 1 gateway server for 100-150 users. In case if the server will be used only as a Gateway without any other RAS components installed then 2 Gateways should be enough.
As for terminal servers there is clearly a matter of expected load, based on the items you would like to publish. So in case if you will publish "large", CPU consuming apps it would be worth having somewhat like 8-10 terminal servers. If published items will consume not that much CPU then 5-6 terminal servers may be enough.

So basically it's only up to you to decide how much GW and TS should you have within the environment based on specifics of future environment. In any ways if you would spot something like overload during peak hours you can always easily enroll new servers to your environment to achieve advanced performance.

I hope that this information proved helpful. If you will have any other questions feel free to drop it here and I'll be glad to provide you with most relevant answers.