Hi Sose, the requirements sound reasonable and standard: Here's a dab at the SLAs that you should be aiming for. I have put them down against each of your requirement for a clear picture. So here they are:
A university portal that will accommodate like 15000 students at once during registration period
# Availability of the website - 95%. If there is any downtime. the web must represent that with an estimated time of recovery. Better still if the website comes up with a timer which shows the countdown when it will be up and running again.
# Full UAT and QA done on the website as per your requirement of the website itself. You should be satisfied with this site before you give them the go ahead to deploy it
Internet connection for 10,000 users, maybe 5 MB bandwidth will do.
# Agree on a QoS. Do you want approx 5 Mb bandwidth but 100% availability, or do you want the full 5 Mb but reduced availability? Like something in the range of 85 - 90%. This is your call. Now if you ask me, I would go to approx 5 Mb bandwidth but 100% availability so that you can manage the rush of students signing up. Yes it will be slow at times, but still you will have web presence 24/7. My reason for this is because, these are students. They stay awake insane hours to get stuff done, not like me, an office goer, who's brain will shutdown after 2100 hours. everyday.
# Monitoring of the bandwidth and usage: Get them to monitor the usage of the website, and I don't mean just hits, I mean traffic and even malicious attempts on the website. A report to be presented on a weekly basis on the above in a format you choose. Now what you want to see in that report becomes their KPI (i.e. Key Performance Indicator)
Internet service should be broadcasted wirelessly
I didn't get this bit clearly, so please let me know exactly what you mean by this.
offcourse network should be built on unix
# Unix requirements: Write down what exactly you need in your unix network, in terms of robustness and reliance. Think in this way, what would you need or think is a good idea for robustness and reliability? Now don't build a ivory tower in your mind. Be practical, if possible be pragmatic. Also consider the people who are going to do this and their capabilities. Do they fit the bill, if not can they procure the knowledge or know how to actually do it? Answer these question, ask them if you need any answers. That will build your Unix requirements. Once that is done, lay it down in front of them as the SLA
HP proliant server
Ok, loads to talk about on this one :
# Which Server, what specifications, what OS, what peripherals, and when I say specifications I mean what speed of processors, what size of HDD, how much RAM, how many network cards, how many partitions in the HDD, will the HDDs be hot - swappable (I would go for this). This is a server. Ensure they maintain this. Agree on a maintenance structure i.e what to maintain, and then once satisfied, agree on a maintenance schedule. Ask the outcome of the maintenance schedule as a report in the same frequency which matches that of the schedule. This is another of their KPIs.
REMEMBER THESE ARE JUST GUIDELINES AND SHOULD NOT BE TREATED AS THE BE ALL AND END ALL SLA GUIDE.
Always remember, be practical, don't ask for something which is fancy but has little usage, but don't ignore something which is vital to the service efficiency of your website and it's associated web based services.
Please, please, please make sure they don't over commit and then under deliver. That will put you in a very bad position.
Ask these people a million questions, don't be shy, because you are going to spend money and time on this.
And lastly for now, don't try to come up with the SLA guide all by yourself. Get some friends and do some research.
Hope this helps for now.
Cheers
Picking pebbles on the shore of the networking ocean
