Also have same issue with QOS disabled on 8.200. Running virtual appliance on VMWare vShere 4.1. Any help would be great..
QoS is totally botched in 8.201 so I've just reinstalled 8.102, upped to 8.103, restored from backup and am back in business...
QoS in 8.103 is a breezeeeeeeeeeeeeeeeee
QoS is totally botched in 8.201 so I've just reinstalled 8.102, upped to 8.103, restored from backup and am back in business...
QoS in 8.103 is a breezeeeeeeeeeeeeeeeee
Actually the QoS in 8.2 is way advanced compared to 8.1. For example you can throttle http with application control and it throttles all traffic including proxy traffic. Throttling proxies with QoS wasn't that simple before this version.
There are some lingering bugs in this version related to http proxy otherwise it's definitely a step forward if they can sort it out.
Hello All
I do not believe, that QoS is messed up in 8.2x. There was changes as mentioned, that it also works with proxies.
I personally would expect, that most people who are experiencing such problems with their migrated 8.1x QoS settings to 8.2 is, that it probably was misconfigured before, and therefor didn't work at all in the past (so it also could not mess up your performance)
Most misconfigurations I have seen in the past (and I have seen a lot of them) was, that they wanted to make QoS for one of the Astaro proxied Services (example Mail Proxy), and created a traffic selector "internal Mailserver - Mailservices (SMTP and SMTPS) - ANY. This traffic selector never matched in 8.1x, because the proxy sends its mails out of WAN Interface (Source for traffic selector is WAN Interface, not internal Mailserver). With Proxies you had to use the ANY object as Source (or your WAN Interface) to create a working traffic selector rule...With 8.2x this seems to be different, because QoS seems to be aware of the original internal IP - also with proxies services (But I'm not 100% sure about this "internal IP awareness in 8.2,...maybe a developer may enlight us to this question...)
So with 8.2 such traffic selectors started to match, and those in the past not working QoS rules started to work as expected...
Only a assumption from me...
BTW: Using Download Equalizer on a WAN Interface is a bad idea at all IMO. From the two used techniques SFQ and RED the second one isn't work conserving. This means it will start to drop packets of the most consuming streams if bandwidth gets saturated and the waiting queue starts growing. This means furthermore, that you will get retransmissions and will saturate your WAN uplink even more...this will automatically end in a crappy performance under load.
For WAN Side only upload optimizer may be recommended in most cases...and it's also working nice in most cases without having to play around with traffic selectors and bandwidth pools [:)]
