Scaling experiences

[technopak]

Hi,

I'm considering offering my Aware app deployed on the web for multiple users using different published BSVs.

Has anyone any experience of this ? Is there any information available about the recommended Server specification, numbers of users supported per Server, numbers of BSVs supported per Server, performance issues etc ?

I know this is a bit like asking 'how long is a piece of string', but I'd be glad to hear any experiences from anyone who is using Aware in a similar fashion.

Thanks in advance.

Peter

[ninap]

hi, good luck getting any response.

we have (other colleagues of mine) posted the same exact questions however not many people share this info.

if you would like we would be happy to share data as we add seats.

also we have looked at scaling based on the layers Web Frontend, Application and Database. Tried doing some clustering of MySQL and also Tomcat but do to a lack of "Testing Software" which would allow us to create traffic based on named users we have no real baseline yet.

pm me and I can get you in touch with our helpdesk to email you some notes so far to date.

txs, nina

[technopak]

Thanks Nina,

I'm definitely interested in sharing information - it's the only way to learn !

It is a bit strange that an application which exists to produce web applications lacks information/recommendations central to that purpose.

Perhaps support can give us a few hints ??


Peter

[aware_support]

This is a bit like asking how many users the 4Ghz PC supports. The answer is it depends. And the list of factors that it depends on is so big, there is no point answering the question in the first place. First and foremost it depends on your application, its structure, rules etc.

[technopak]

Thanks support,

That's pretty much as I expected.

We'll proceed on the basis of 'Let's try it and see'


Peter

[technopak]

Just a quick last question or two -

Assuming adequate memory and processor speed, does the OS where Aware is installed affect the application performance ? Or once Aware is running does it control the speed etc.

Is there a recommended OS for web deployment of Aware apps ?

And also, is there a recommended ratio of memory allocation - JBoss, Aware, Tomcat ? If I have 3GB free, where should I allocate it to get the most benefit ?


Peter

[ninap]

peter,

re: memory allotment:

you may run into the same issue we have with larger allotments or +2048 per layer. when you shut down JBoss, and restart you may find it does not
restart using the console to a memory fault.

this is something our helpdesk is reviewing but the work around is

workaround: startup with c:\awareim\bin\startAwareIMc.bat
chage the memory back to 1024 layer/app and then startup with console panel "or the normal" desktop icon

performance: we have viewed any change between deployments on Apple/Linux or Windows XP/7/2003 or 2008 - But this was an "eyeball"
and not diagnosticly. but as we add users we will let you and the forum know.

txs, nina

[NBDAB]

Peter, another item which Nina should mention is disparity between server size.

Our helpdesk or infra group was initially thinking about Aware more like an enterprise application farm. And after months of testing in many different Servers, OS's, Layers we have found that it really is not as necessary until your reaching a user threshold per box, or connection.

As support mentioned. The variables to consider are many and at each point could be a cause of latency or concern. However for the most part its "Throughput" from DB to Application to Web Frontend to Internet to Customer to the customers Environment"

When you consider the multitude of typical customer "desktop" issues and there consumer internet services you will find your fighting a possible battle of fine-tuning which is overkill.

So our current principles follow the SaaS standards of APDEX

http://apdex.org/overview.html

We look only at the delivery from our DC to the customer’s location and that latency is resulted in the APDEX score. We use this as our SLA towards the customer and now fine tune based on this.

You need diagnostic software at the DC level and a good understanding of each layer to help troubleshoot and diagnose latency or errors. Following keys like "Low UI or GFX", "Limited View Perspective Queries" and a couple other of our own internal guidelines have allowed us to get a Rich user experience on lower cost cloud equipment.

Note:
We have tested on 512 VD's (Virtual Desktops) to now 16gb mem, 8 core servers (non-virtualized or bare metal)

Scaling:

You can take Aware vertical or horizontal. Meaning when broken down
Tomcat scales horizontally best and with most bang for the buck
JBoss we stick with vertical as Bigger boxes are cheaper these days
MySQL cluster (beign tested now) offers enough Speed and Catastrophe/Safety for High Availability SLA's

Thanks,
Lars

[aware_support]

Some additional words about JBoss/Tomcat/Aware IM server memory ratio. JBoss will probably be happy with the default setting (64Mb) unless you have really large chunks of data transferred across the network.

Tomcat should be bumped up from the default setting of 64Mb when you have many users accessing the system concurrently. Most memory should be allocated to the Aware IM Server - the more the better. 256 or 512 Mb will probably be more than enough for the vast majourity of applications.

The choice of OS shouldn't matter much as far as performance is concerned.

[technopak]

Thank you everyone,

The information from these contributions is very useful.

Any more out there ??


Peter

[rocketman]

Hi Pete,

This response isn't as technical as the others - just some real world numbers.

Hardware - 2 x quad core sempron 2.4Ghz processors with 2Gb Ram

Jboss = 128Mb
Tomcat = 128Mb
AwareIM = 512Mb

Server is also running IIS6 for standard web deployements

BSV1 = 37 BO's, 110 processes - 189 users of which up to 30 active at any one time

BSV2 = identical to BSV1 It's a learning centre where members can learn how to use the main system without screwing up the data Max recorded users at any one tim = 15

No reported issue with speed or response except at our launch control vehicle where because of the surrounding terrain and our location in the (UK) Derbyshire Peak District we have to a) Bounce our wi-fi signal through two repeaters to end up at a really crappy broadband connection (1.5Mb/sec download and 150k upload on a good day) Even then, the system is very useable (actually now in active daily use from 09:30 to 19:00) At the end of the day it can take about 6-8 secs to refresh 50 active flight records - which we put down to the broadband. The same refresh on an 8Mb link takes 1-2 secs

So in commercial terms - not a big app - just thought it might help

[technopak]

Rocketman,

Real world feedback much appreciated

I know there are so many variables that it's hard to be scientific in analysing possible system performance, but your anecdotal experience is really good to hear.

My app would be similar in structure to what you describe but I will most likely have a separate BSV for each group of users (at different companies etc). I'm wondering how many BSVs will run happily on the same Server - or maybe the number of BSVs doesn't matter ?

So far, I have only a small number of users (max 30) on the system - and it's performing well on very modest hardware -

Shared VPS with quad-core 2.5 processor, 1 GB RAM.

Scaling up is going to be interesting to say the least !

Thanks for the post,

Peter

[rocketman]

Hi Peter (is it Peter or Pete BTW)

My suggestion would be to beef up your memory. I don't think it matters how many BSV's you have It's what the rule engine can handle that counts (I think - perhaps support can confirm)

When I've on occasions opened up the Aware viewer I've noticed all the process actions etc from all sources - be they from published BSV's or Under Test BSV's all get passed to the same engine. IE there aren't multiple engines running - one for each BSV. At one time I had a couple of independent apps running to test some logic before integrating them into the main app and they all ended up in the same place

Hope that make sense

Mo