Quality Center / ALM Practitioners Forum
cancel

Bandwidth Utilization on Average of ALM 12 desktop clietn

BrianPalagyi
Respected Contributor.

Bandwidth Utilization on Average of ALM 12 desktop clietn

I have heard this question from a number of clients, and I am a little surprised by it to be honest.   but I am wondering if anyone out there has seen or recorded a range of different network latency speeds, documented the bandwidth utilization, and response metrics for those solutions. 

5 REPLIES
Damodar-Reddy
Acclaimed Contributor.

Re: Bandwidth Utilization on Average of ALM 12 desktop clietn

Brian,

 

It's not the same allways, even though you recorded the time responses, Since it depends upon the various verticals of the Considerations.

1. Network Speed (Bandwidth)

2. RAM of the Local / End Client (May be VDI)

3. Application on WEB or APP?

4. Server Pre-requisites ( Heap Size).

 

I can understand this could be litle slow, if you can compare to the previous versions of this.

Since much Indulged in Performance & features....FAIR to accept some delay.

 

 

(If this helped, accept as solution )
(Posts and opinions made here are my own and do not reflect the opinions of my employer)
Craig_Drummond
Outstanding Contributor.

Re: Bandwidth Utilization on Average of ALM 12 desktop clietn

Damus,

 

Your answer only addresses response times and doesn't address the heart of the question, latency and bandwidth. 

 

Latency is the delay due to distance from servers or slower network speeds and bandwidth is the amount of data that is being sent from the server to the client.  While there is certainly a difference in bandwidth between the web client and the ALM client, the amount of memory on the server or the client and the network speed have absolutely no impact on how much data the server has to send, it only effects how fast that data gets to the client.

 

My understanding of Brian's question is,  as part of the testing that we do to create the Benchmark document (the internal performance testing of ALM) are there any findings or recommendations that can be given to customers regarding latency and bandwidth?  In other words, have we tested ALM using several different bandwidths and with clients (ALM Clients) from many different locations and with different latencies.

 

Also, if you were to conduct a performance test of ALM 11.5, 11.52, and 12 are there differences in the total amount of data that is being transmitted over the network that could make latency and bandwidth issues more pronounced?

 

Craig

BrianPalagyi
Respected Contributor.

Re: Bandwidth Utilization on Average of ALM 12 desktop clietn

Correct Craig.  I understand the mechanics of tuning and test data validation between the differetn configurations and set ups, I am looking to see if we have metrics for latency of X vs. latency of Y on a given environment.  the closest data i could pull was an old Mercury QC Version 8.2 document that had a breakdown of latencey figures from 20ms to 200ms and hops of 2 to 5 within a local vs. internationl sessions where it showed the repsonse metrics were 'kinda' stable until you hit the 100ms mark, where the repsone metrics hockey sticked greatly.  but I was wondering if any other data on network speed or future latency discussions/validaiton had been done by anyone out in the field.

Damodar-Reddy
Acclaimed Contributor.

Re: Bandwidth Utilization on Average of ALM 12 desktop clietn

Well Craig,

 

That's what I mentioned Overall, Latency, Bandwidth, Performance,Speed of Upload/Download in to QC. Cannot be Simillar in all the environments.

Will vary if you consider / Compare two versions (Previous VS Immediate).

 

A simple Concept of Knowledge, You cannot fit all your books in your Old School bag when your Standard moves on. you need a new school bag, which covers & fits your requirement. Just like New version of application needs some more Rather than the old and Performance will obviously Observed (LIL DOWN)...Whether in Latency, Bandwidth...or what ever the terminolgy is..!!

 

By the way Heart of the question is any one Documented these..??

(If this helped, accept as solution )
(Posts and opinions made here are my own and do not reflect the opinions of my employer)
Highlighted
Craig_Drummond
Outstanding Contributor.

Re: Bandwidth Utilization on Average of ALM 12 desktop clietn

Damus,

 

I re-read my original reply and I don't see anywhere that I asked if the current version has the same size data transfer as the old version.  What I did asked is if we have documented the difference and if that difference could be affecting customers who have small bandwidth or large latency issues on their network. 

 

In your reply you suggest that the amount of memory on the server, the amount of bandwidth available, or the latency between the server and the client would some how effect throughput which is not correct.  An ALM server with 8GB of RAM has to send the same amount of data (Throughput) as an ALM server with 32GB of RAM.  And that same amount of data has to eventually get to the client regardless of your connection speed or distance.  The only thing that these factors determine is response time, something that was not mentioned at all in the original question.

 

Brian's point is that he found a benchmark document from an 8.x version that discussed throughput, but subsequent versions of the benchmark document have for some reason not reported on throughput.  Why is that? 

 

It is not a loaded question, it is just a simple fact that the more features and bells and whistles that are offered in the software there will absolutely be more throughput required.  But understanding the degree of increase will tell us if our book bag (the network) needs to be increased by a second book bag, or do we need a wagon, or a whole new book shelf. 

 

Not reporting the throughput doesn't mean that we have not caused a problem, especially for those who have customers in far off locations with low bandwidth and high latency.  Knowing if ALM requires a minimum of 100mb network connection or no more than 200ms of latency is a valid concern, especially if you have your ALM servers in Chicago but you have teams accessing that server from Japan or India.

 

Can someone who has worked on the load tests that were executed to create the ALM Benchmark documentation please weigh in with the actual data that was collected regarding throughput for ALM 11, 11.5, 11.52, and 12 so that way we may determine for ourselves if the level of throughput increase is a cause for concern.

 

Regards,


Craig Drummond

HP Premier Support Technical Account Manager - ALM Products