savage
Expert Member
It also affects much more than just MYSQL. Any application that is sensitive to fragmentation will suffer.-
Join the MyBroadband community
South Africa’s biggest forum. Discuss, discover, and connect with thousands of members.
-
Question of the Day: Do you think you could pass a driver's licence test without any preparation?
Afrihost Business Uncapped Feedback - 2
- Thread starter The_Unbeliever
- Start date
- Status
It also affects much more than just MYSQL. Any application that is sensitive to fragmentation will suffer.
I don't think we're intentionally degrading or shaping services on Business. We've made changes to make traffic management more efficient. If some services are performing poorly, it's due to overall contention, which we're trying to manage as best we can during the peak periods.
cortexenslaver
Senior Member
Where in Cape Town are you situated?
I have found recently that Telkom is also upgrading various areas, so the latency could be an exchange issue. I am in Sunningdale, Table View, and there have been some really terrible speeds on ALL ISP's lately.... and then, like magic, all is well again. Sometimes it's really hard to know (without a lot of continuous testing) whether it's your ISP, or Telkom line.
That said, when all other ISP accounts work flawlessly, AH accounts are still abysmal
MySQL traffic shouldn't be affected on an unshaped product, but we have had some problems with Static IPs (if you are using one on Business). Otherwise we'll need to investigate this and see what could be going wrong.
I see in a later post you mentioned working with Critical Care. Can you PM me so I can follow that up for you?
It is very difficult to predict user behaviour on a massive scale. Sometimes usage spikes inexplicably and then we have to try to compensate in the short term while we accelerate our long term plans. As an ISP we have to constantly balance the cost factor of having too much unused capacity, which affects end user pricing, and having too little which offers a poor experience. When you factor in the long lead times for ordering capacity from Telkom, this is always a very delicate process
cortexenslaver
Senior Member
The main problem I noticed with AH support, is that it doesn't seem like techs/reps/agents have access to the history of a ticket. Their system shows an open ticket, and you may get 5 different people responding. I have had a ticket with critical care for 2 weeks before... in the mean time, I had another 7 people respond to it, giving me the same "generic responses" of, "can you do a trace route, etc."... this after I had submitted them on 3 previous occasions, and after the ticket had been open for almost 2 months.
It cannot be that hard to have a ticket system that keeps track of client and rep/agent responses, can it?
I have 4 different providers (2 in SA and 2 in the US)... all use different ticket systems, but all have the same thing in common: TICKET HISTORY + CLIENT INFO/HISTORY.
Last edited:
Definitely doesn't look great there. Is your general throughput fine? on speedtests or other services?
Thanks for the faith and trust you're extending here. Believe me, it really means a lot. I do believe we've always made it work for our clients in the past and we'll continue to do so in the future. We'll definitely keep trying on our side until we feel our clients are getting the performance they should.
MrGray
Honorary Master
Sure, but I can honestly say that in over a decade of working remotely on MySQL servers, I have never had a situation where queries randomly timeout due to issues at an ISP and MySQL has always been quite robust in this regard.
Our network status is always weighed up based on the overall impact on the network. So while this is definitely a critical issue for many clients, the large majority of clients are connected and getting adequate performance for the business, personal use, etc. I don't think that the label should be any indication of the priority this issue has and the level of commitment from our side to resolve this quickly and to the benefit of our clients.
MrGray
Honorary Master
That is exactly what's happening to us. It's bizarre. You explain the problem in detail to one person, meanwhile some other random person sends an email asking you if tried rebooting your router.
Interesting that both accounts are seeing latency spikes, I'm not sure how you are measuring this because the tests you've posted are different. Are you seeing this on traceroutes or ping tests? If you can show me where on our trace the latency spikes we can definitely look further into that
4saken
Active Member
Speedtests are not relevant here. Its got nothing to do with my Telkom line. I switch to a Plugg account and it goes at 1000KB/s. My telkom line is perfect. This is specifically Afrihost related. But it also shows that everything is in fact being shaped, I can't even download steam games at full speed anymore.
Both tests were straight forward DOS prompt pings (500 pings).
AfriMan, I asked yesterday whether using a capped unshaped account would be a viable interim solution?? Any response?
Could you possibly supply me with a test account to do some testing with??
AfriMan, I asked yesterday whether using a capped unshaped account would be a viable interim solution?? Any response?
Could you possibly supply me with a test account to do some testing with??
^ This
AfriGuy/Man, can you please give this feedback to the relevant people at Afrihost as well. It does make it difficult if you have to start from scratch every time. (For any enquiry)
savage
Expert Member
Wow really? you call this adequate performance? It's no wonder nothing gets done about it yes!
I wasn't implying that your line is at fault. I was wanting to know if you were getting general poor throughput on your account. Everything being shaped is throttling, and we don't do either, so we need to see if your account is just generally performing badly.
Business and Capped accounts are constructed from the same "data pool" so performance should be very similar. I don't think that changing would have any effect, unless you were looking at an account without a product rate limit or more concurrent connections.
I think we need to look at exactly which services are not performing and try to improve them wherever we can, or look at improving your overall throughput or latency, whichever is affecting your performance.
Whenever we escalate a query here from MyBB we try to include the history of the conversations we've had.
In terms of support, all ticket history is saved and each consultant should have full view of the history, as well as notes made on the individual account/profile. We definitely have the systems in place, but perhaps our people are (in some cases) not using the tools at their disposal.
If you come across this, please PM me so our QA team can identify these guys and we can give them more support and training
Crusader
Executive Member
Afriman, any comment on those Diablo 3 pathpings?
- Status