A couple of things I would like to see in a book (and seems to have never
been covered before):
1. How to structure large servlet applications (Data model).
2. Scalable/fail safe servlet applications.
i.e. Our application is accessing a database of up to 100Gb and we cache all
of our data in the memory of the VM rather than reading the DB for every
request. The VM can only handle 2Gb, but in reality should hold 500Mb or
less. The data model is very complex and is very sophisticated. We have
features such as lazy writing where it updates the record in memory, but
doesn't send the changes to the database immediately; early reading where we
do a big SQL select in replacement of lots of small SQL selects and cache
the result set until ready to build our internal data model.
We are less concerned with performance in our application because the real
limit is the amount of memory the VM is currently using (50 users are likely
to be using 250Mb), therefore concurrent users aren't going to cause a large
amount of performance issues, but could potentially cause memory problems.
As the data is loaded in several separate VM's fail safety became a problem,
but has been sufficiently resolved for our app. Also data that needs to be
shared between instances of the VM is very difficult. We have had to
broadcast to the other VM's that a particaular record is now invalid and the
next time it is used it goes back to the DB.
We have had consultancy from IBM on these issues and we wanted to find
another company and books to help us as well, but we have failed so far. By
the way if anyone knows of a consultancy company that has had experience of
projects like this please forward their details to me.
If you want to discuss the issues we have resolved and how we have done it;
or discuss the issues that still remain EMail me.
Dave.
[EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, email: [EMAIL PROTECTED]
