Using Session to store Datatable

Question

currently we are using Sessions to store datatables in our pages so that we don't have to make Database hits to get the same datatable again and again. But my worry is that it is using the server memory and if large number of users login some day, the response of the server will become slow and our application might crash as well.

Please tell me is it a good idea to store datatables into Sessions or should we get the datatables from DB everytime?

Answer 1

As a general rule of thumb I would say don't use session. I haven't had to use session for a long time. As soon as you move into a web farm situation session either gets a lot slower or a lot more complicated or both.

Whether you will get away with it or not really depends on how much data you are storing in session, and how many users will be active within the session timeout period.

There are a lot of caching and in memory database options available today that may be a better option. Finally, while the solution as described sounds questionable, I wouldn't optimize the existing solution until you have actually measured a problem.

Answer 2

This is dependent on what is being stored in the datatables. In any case, I would use the ASP.NET Cache to store these datatables for the following reasons.

• Cache has an expiry, which means you can automatically remove it based upon a sliding or absolute expiry timed value

• Cache will automatically be removed if the processes memory "pressure" is too high.

• You can make a cached item specific to one user, or global to all users based upon its key

for example:

// personalized cache item
string personalCacheKey = string.Format("MyDataTable_{0}", (int)Session["UserID"]);
DataTable myPersonalDataTable = (DataTable)Cache[personalCacheKey];

if (myPersonalDataTable == null)
{
    myPersonalDataTable = database.dosomething();
    Cache.Insert(personalCacheKey, myPersonalDataTable, null, Cache.NoAbsoluteExpiration, new TimeSpan(0, 30, 0)); // 30 minutes
}

// global (non user specific) cached item
string globalCacheKey = "MyDataTable";
DataTable globalDataTable = (DataTable)Cache[globalCacheKey];

if (globalDataTable == null)
{
    globalDataTable = database.dosomething();
    Cache.Insert(globalCacheKey, globalDataTable, null, Cache.NoAbsoluteExpiration, new TimeSpan(0, 30, 0)); // 30 minutes (again)
}

The issue that you have now, however, is if the underlying data gets updated, and whether it is acceptable for your application to present "old" cached data. If it is not acceptable, you will have to forcibly remove an item from cache, there are a few mechanisms for that.

You can setup a SqlCacheDependency (which I have never personally used), or you can just clear out the cached object yourself using Cache.Remove(cachekey) .

Answer 3

It is preferable to store "commonly used" data in Memory; that's good logic. However "Session" means that it exists for the life of that Session, and hence that user. Secondly, pending the user "Session" life, as you already said, this could be consume valuable resources on the Server Side.

What you may want to consider using is the "Cache" object, as it serves the same purpose with "Expiration".

DataTable users = new DataTable();

if (Cache["users"] == null)
{
    // users = getUsers(customer);
   Cache.Add(“users”, users, null, System.Web.Caching.Cache.NoAbsoluteExpiration, new TimeSpan(0, 60, 0), System.Web.Caching.CacheItemPriority.Default, null);
}
 else
{
    sers = (DataTable)Cache["users"];
}

There are many ways to re-use memory in .NET (1) ViewState (2) Cache (3) Session (4) Cookies

But I would go for the "Cache" object.

Answer 4

If you can't increase memory on the web server then the obvious answer is to not store it in session state and get it from the database every time.

The problem with this is what impact will it have on your database? Are you just moving the problem from the web server to the database server?

It is much easier to scale out web servers than it is to scale up/out Databases (and often cheaper if you're using something like SQL Server)

Answer 5

如果您的数据表具有较少数量的记录并且它不包含敏感数据，那么您也可以使用ViewState但数据应该更小，因为此方法将序列化数据并将其存储在客户端，然后从客户端获取数据到存储在服务器端。