Orion Circuit Upload Speed Requirements

Hey everyone,

We have an HA Orion environment configured for a rather large network, but one of the issues we've noticed is web server performance on our Additional Web Servers. We only have a few engineers using the platform so far, so we know it's not a load balancing issue, so our thoughts turned to the Comcast circuit we have serving the environment.

Currently, the circuit is performing at 175Mbps down and 25 Mbps up. Is this sufficient in anyone's experience here for a well-performing Orion web server? Or should we be looking at more symmetrical connection options?

Thanks!

Ted

Find more posts tagged with

Performance

network

bandwidth

npm

awe

server

web

Accepted answers

cnorborg

For using Solarwinds via the web interface, or for polling remote sites, the bandwidth you stated should be more than sufficient, assuming no other major loads are on it. The HTTP pages and SNMP polling shouldn't generate that much traffic at any given time. I should think even if it was a symmetric connection at the lower of the rates mentioned you should have plenty of bandwidth. Since you didn't mention otherwise, I'm assuming that all the sites are directly connected to the Internet and there is no VPN or tunnel happening between them? Maybe perimeter firewalls to restrict what traffic gets in or out? A VPN could add additional overhead, esp. if it wasn't set up the best and was fragmenting a lot of traffic.

If you are having performance issues, the first thing to check is how you have your database set up. This is the most critical component because if it isn't running at peak performance, you will see this cascade to your poller and AWS's. The poller runs a close second, but it tends to run ok if the database server is running quickly. If you don't already have your main box / poller and the database server on different boxes, make sure you do this! Be sure to allocate plenty of memory and CPU power to the database server, with dedicated hardware being preferable over a VM. Be sure that you don't have things like mirroring your drives turned on. SSD's are always preferable over spinning disks... The modules you have can make a difference too, like with NTA for peak performance have it on a different SQL server also, but that depends on your utilization and such.

All comments

cnorborg

A bit confused. You have HA set up, is it across the comcast link you set up? Could you describe your environment a bit more? I'd assume your main server (not the HA) is in the same location as the database server? Where are the AWS's also?

For running web pages over a 175/25 link, I don't think there should be much of an issue. But that depends on what else is on the link. Running a database connection from either a primary or HA server back to a database across a link might be, and remember that if you have an AWS sitting across this link, if you're local and connecting to a web server across the link you have increased effects from latency. ie: not only do you have to download pages from the server across the link, but the AWS needs to query the database over the same link I'd assume.

tsadler

@cnorborgThanks for the reply. Sorry for the lack of description -- here are some more details

I was mistaken about our HA configuration. Spoke to our system architect, we have our AWS's in load balancing configuration but there are no other redundancies in place.
There is only one incoming ISP connection, the Comcast circuit I previously described
The entire environment is situated across two Hyper-V hosts, so all machines have a direct access path to one another

So, the only traffic traversing the link is for web services or remote polling. We have a lot of remote infrastructure locations, but most of those have not been added into NPM yet for monitoring.

cnorborg

tsadler

Thanks for the input! I'll relay this info to our system architect and see what other bottlenecks could be causing this problem.

Much appreciated!