A close look into HCL Software download performance
Daniel Nashed 14 October 2023 08:12:06
When looking into Domino V14 EA3 autoupdate.nsf downloads I have been testing some proxy connections with different locally installed proxies (Squid HTTP proxy and Dante Socks proxy).
On first look the performance seemed to be dramatically different.
But on second look the difference was the way I tested the download. The second download via proxy was almost as fast as the performance I had without a proxy.
When testing performance, caching can dramatically influence the results. Usually it is disk cache performance where you have to drop file caches on your machine.
My HCL Software download performance depends on server side cache
In this case it turned out that the My HCL Software download caches the files in the Content Delivery Network (CDN).
If nobody from my region downloaded the file for a while, the file needs to be read from the back-end.
Testing with different network connections
Of course download performance also depends on the internet connection. My AWS machine was surprisingly a lot of slower than my Hetzner server...
Looks like they only have giga bit connections for their normal servers. But then 114 MB/sec is as good as it can be!
For a first download without the cache I got like 15 MB/sec on a AWS machine and 115 MB/sec for a cached file.
On my Hetzner server I got more than double the performance. Once the file was cached I hit a download rate of 365 MB/sec!
Another interesting surprise was that my ARM machine was 20 MB/sec faster than the Intel machine (Intel Xeon Processor, Skylake, IBRS, 3300 CPU MHz).
And 50 MB/sec faster than a machine with just 1 Intel CPU core.
All in all the performance for downloads from the new My HCL Software Download site rocks!
To be fair I did the same test with Flexnet. I got some mixed results. But they also cache the download in some way.
The first read was half the speed I got for a none cached file on the new portal. But I got comparable download speed on second download.
Conclusion
OK those are not the number you can expect in normal corporate downloads at all. But it shows which factors influence download performance.
If you have a proxy which inspects downloaded data, the performance could also look completely differenct.
But my test was mainly a check if proxy connections works well and to see where different download performance came from.
It's still a pretty impressive download performance.
-- Daniel
Test result with curl command line (no proxy) on a Hetzner machine with ARM CPU (Data center located in Germany)
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 1052M 100 1052M 0 0 365M 0 0:00:02 0:00:02 --:--:-- 365M
- Comments [0]
![[HCL Domino]](../lotus-domino.gif){kind=small}
![[Domino on Linux]](../domino-linux.gif){kind=small}
![[Nash!Com]](../nashcom.gif){kind=small}
![[Daniel Nashed]](../daniel-nsh.gif){kind=small}