New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Required Dedicated Server for the Online Education Exam Software to handle 5K concurrent users
MiteshGanatra
Member
in General
Required Dedicated Server for the Online Education Exam Software to handle 5K concurrent users with 100% uptime and server won't be crashed guarantee.
Comments
What is the software name? What is the software server requirements? How many resources needed per 50/100/200 users? Please provide as much information as you can, because I am 100% sure that no one in this forum will not be able to tell you what server parameters is needed for 100% uptime without a crash. We do not want to do magic tricks.
What's location are you looking for and what's your budget ?
This is custom made software running on PHP and MySQL. Currently we're having a server with GoDaddy and the configuration is 4 GB RAM, 1 TB HDD, 4 Cores and Unmetered Bandwidth. What is happening with their server is once the 400 users are connected to give online exams, server don't crash but the next user no. 401 is not connecting. They provide the 100% server uptime guarantee, but unfortunately not able to handle more users.
This is mainly for the Indian Colleges and Universities, I think the Asian server would be preferable, but US/UK data center won't be any issue cause we are planning to use CloudFlare CDN. Also, there is NO sure budget for the requirements. All we want is 100% uptime and server won't be crashed guarantee.
First of all, no one in this work can't guarantee 100% uptime. If they telling that they do - they simply lie.
I recommend to increase RAM and change HDD's to SSD's in RAID10. Not sure about CPU loads, you can tell us more of you manage this server.
It's hard to tell what the issue might be. Do you receive any HTTP error messages, or does it time-out? Also, at that time, are you out of resources? (i.e RAM/SWAP/etc)
Perhaps you can run a load test and monitor how the resources get used (i.e try loader.io) so that you can estimate the resources you would need. We have an X3430 server with 16 GB and 2x1 TB HDD in stock, and we can provide up to 128 GB of RAM on higher CPUs (i.e in case NL is fine) if you think you might need more resources.
India, go with @LeapSwitch.
Singapore (dedi is better for the purpose?):
LeaseWeb
NewMedia Express
You can try Ctrls.com. They are pretty solid.
for cheaper options try XINIX.IN or LeapSwitch.com
looks like connection limit to me. did you check the logs?
We offer Managed Services to a lot of clients with similar levels of traffic in our India and USA locations. We can also do load balancing across multiple servers, MySQL Galera clusters, GlusterFS file replication etc. You can reach me on Skype - leapswitch , or Email - ishan at leapswitch.com to discuss your requirements in detail.
Database connection limit, try distribution across multiple vps. This will allow you your "100%" uptime, which is allmost impossible to do with a single server.
If this is a product and you have budgeted for success, then perhaps google appengine will help you with scaling properly.
this sounds more like your custom software wrecking hard than a server issue
Problem probably exists between HTTP connection and the hardware. -> your software stack needs tuning first.
What is your RAM budget per user?
Do some profiling for your PHP loads and actions with 10, then 50, then 100 test users doing real clicks with random timers.
When in doubt throw more hardware at the problem.
I am guessing, You only need these resources for a short period.
Talk to DigitalOcean Bangalore about spinning up a bunch of PHP frontends and a large RAM database VM at the particular pre-arranged time.
Try some or all of those random suggestions that are based on knowing very little about your setup.
You would want to have a load-balancing servers. Load Balancers manage online traffic by distributing workloads across multiple servers and resources on demand. They maximize your workload performance and help prevent overload to help give your users a seamless experience. I would recommend HAProxy configuration with at least 2 servers to accommodate it.
We can offer a good dedicated server in US. You may please look at https://thestack.net/site/serverla
Meantime, you can simply mail us your exact specification so we can discuss it further. Our mail address is [email protected]
Try Digitalocean Indian location.
No single server is 100% uptime guarantee.
Big companies offer good SLA, eg. AWS offers 99.95% uptime SLA on their EC2. 100% uptime is basically a joke.
I see, however 100% uptime SLA will work out and numerous organizations guarantee to give what it is mentioned in their SLA.
We ran bunches of tests and found there was server side issues, they ensure their servers in some way or another and don't reveal why they are doing as such. Thus, searching for new provider.
Will consider, thanks for your suggestion.
Will consider, thanks for your suggestion.
Did that mate and they put the limit to protect their servers. They ensure their servers in some way or another and don't reveal why they are doing as such.
Make sure you understand what an SLA is. It doesn't guarantee up time, it's an agreement that you will be reimbursed for the cost of your services during the time they are down.
Make special note that you will not be reimbursed for loss of business or inconvenience. If you pay a provider $30/mo and your services are down for 1 day, you'll get a $1 minus the reasonable repair time clause.
This whole thing sounds like a train wreck, but . . .
We did every one of the tests and discovered they set the points of limits of their servers. We expanded the concurrent limits from apache yet did not fill in as there were impacts reason for server side.
Thanks mate. That's what I was looking for to load-balance the servers beginning with three 8 Gigs of servers and will expand the RAMs as it requires in the future. I just PMed you, please revert me back with your suggestions to consider.
If you are looking still, we can offer:
SQL Server (Shared) :
16 GB RAM
4x512 GB SSD
Front-end:
120 GB Storage Space
Unmetered Bandwidth ( 1 Gbit Port )
99% Uptime Guarantee (backed by SLA)
DDoS Protection (via 3 different filter providers depending on location selected)
Load Balancing (support up to 150,000 requests per second without slowdown)
Akamai CDN w/ Delivery Domain (costs $3,500 one-time fee)
This would require a signed contract, minimum 6 month commitment, and would cost you US$1200 every 6 months, or $2400 a year
you are on a lowend site asking for sla and 100% uptime...
What is your website?
No matter how much raw power you throw at it, there is no way around, except fixing shitty code and tuning / optimizing your stack, which seems to be the issue and the requirement in your case.