Storing ASP.NET session outside webserver – SQL Server vs Redis vs Couchbase

If you are storing ASP.NET session in webserver memory, then when a webserver dies, you lose all the sessions on that server. This results in poor customer experience, as they are logged out in the middle of some journey and has to log back in, losing their state, sometimes resulting in data loss. A common solution to avoid such problem is to store session outside the webserver. You can use a separate cluster of servers where you run some software that lets your application store session. Some popular technologies for storing session out-of-process are SQL Server, Memcached, Redis, NoSQL databases like Couchbase. Here I present some performance test results for an ASP.NET MVC 3 website using SQL Server vs Redis vs Couchbase.

Test setup

Configuration of the servers are as following:

  • Web Server: 2 x Windows 2008 R2 Enterprise x64 Virtual Machine running IIS 7.5, each with 4 x vCPU, 2100 MHz, AMD Opteron Processor 6172
  • Load Balancer: HAProxy on one Linux VM.
  • Load generator: 2 Linux VM running JMeter
  • Test Plan: Perform a Login and then use the session cookie generated to visit Home Page. Ensure home page shows the logged in user’s name.

SQL Server ASPState DB as Session Store

SQL Server 2008 R2 configuration is:

  • CPU MAX SPEED: 2100 MHz
  • CPU NUM: 4
  • CPU VERSION: AMD Opteron(TM) Processor 6172
  • RAM: 8 GB

SQL Server Mode

Now the juicy stuff: I am able to produce 567 requests / sec, while the response time median is 48 ms. Those who think SQL Server is slow, not for in-memory stuff, think again.

SQL Server Result

SQL Server response time

When I restarted website on one webserver, it resulted in a small fraction of error and a spike in response time:

SQL Server restart


Redis runs on two Linux VM. Each running 3 instances. They are Master-Slave clustered. 3 Master, 3 Slave. Configuration of each Linux VM is:

  • CPU: Intel(R) Xeon(R) CPU  E5540  @ 2.53GHz
  • 4 vCPU
  • 8 GB RAM

Redis diagram

Redis load test

Redis gives 707 requests/sec. The response time median is also improved 38 ms. This is about 150 Req/sec more than the SQL Server result.

Redis response time

Notice the periodic jump in response time? I have seen this happen consistently with Redis setup. This must be when Redis decides to take a snapshot of its memory, and dump it to the disk.

Restarting one webserver results in a small fraction of error. However, I have experienced a total outage on the website when a Redis master node was killed, when I was using the Microsoft’s Redis Session State provider implementation. But when I used Alex Simonov’s Redis ASP.NET Provider, which uses StackExchange Redis client library, it was able to immediately failover to other surviving nodes, without causing an outage on the website.

Redis restart


Couchbase is running on 2 x Windows servers:

  • CPU MAX SPEED: 2100 MHz
  • CPU NUM: 4
  • CPU VERSION: AMD Opteron(tm) Processor 6172
  • RAM: 8 GB

Couchbase diagram

Couchbase offers two types of storage, a Memcached-based fully in-memory storage, and a disk-based storage, where data is first written in the memory queue, and then flushed to disk when the queue is full. I have used Memcached in-memory store for the session store.

Load test result looked bad:

Couchbase load test

Couchbase response time

This is slower than SQL Server! However, looking at the code, I see the Session State provider implementation is not as optimal as SQL Server or Redis Session State implementation. Those hit the server once per page hit. But the Couchbase implementation hits 4 times per page hit. It is possible when Couchbase releases a better implementation, it might outperform SQL Server and Redis.

Couchbase dashboard confirms the 4x hit observation. For 400 hits to the webservers per second, it results in more than 1.5K hits to Couchbase.

Couchbase Dashboard


If you want to stay pure Microsoft stack, go for SQL Server-based session. You should setup database mirroring and put the mirrored database on ASP.NET connection string so that when primary SQL Server node goes down, the secondary comes up and the application is able to detect it right away. If you have no issues mixing Linux in the stack, then Redis is an awesome technology. You can also use Microsoft’s compilation of Redis that runs on Windows.

Powerful IIS/Apache Monitoring dashboard using ElasticSearch+Grafana

IIS or Apache do not come with any monitoring dashboard that shows you graphs of requests/sec, response times, slow URLs, failed requests and so on. You need to use external tools to visualize that. ElasticSearch and Grafana are two such tools that let you collect logs from web servers, and then parse, filter, sort, analyze, and create beautiful presentations out of them. ElasticSearch is a distributed JSON document store, just like a NoSQL database. You can use it to store logs as JSON documents. Then you can use Grafana to fetch those documents from ElasticSearch and build beautiful presentations. Both are free and open source. 

Web Graph

Read the full details here, please don’t forget to rate:

ElasticSearch is a very powerful product. It is a multi-purpose distributed JSON document store and also a powerful search engine. Most frequent use cases for ElasticSearch is to create searchable documents, implement auto completion feature, and also aggregate logs and analyze them. Grafana is a beautiful Dashboard tool that takes ElasticSearch, among many, as a data source. Combing these two, you can build sophisticated monitoring and reporting tools to get a holistic view on how your application is performing and where the issues are.


jChaart – Web Dashboard Framework

A Javascript-only, web based Dashboard library, that you can use to make eye-catchy Charts showing Transactions stats for your application, Server CPU, Memory, IO graphs etc. It offers libraries to convert different types of delimited text files into Charts. Those delimited files are generated from various sources, eg running SQL queries against a Database, or running shell scripts to collect system stats, or Powershell scripts to process IIS logs. End result is a nice-looking Twitter Bootstrap powered, responsive Web Dashboard, that you can get up and running in no time, on any platform. Since it is HTML and Javascript, you can customize it to show exactly what you want, how you want. Quite handy for earning brownie points and wooing your customers.

Here’s an example how a Transaction Dashboard may look like:




And an example of a System Monitoring dashboard:


The GitHub project is here:

You are welcome to join and participate.

Student Data Bank: An ASP.NET MVC, WebAPI, EF Codefirst, Task, ESB showcase

Imagine this: you are a student of Oxford and you want to join a program in Cambridge. Cambridge wants to pull all the courses you have done in Oxford, along with your course results and automatically credit them, so that you don’t have to repeat the courses. Eventually when Cambridge decides to award you a degree, they can automatically check that you have done all the courses required by their program, either in Cambridge or in Oxford, Harvard, MIT or any other universities in the world. Would it not be nice to have a common system where all your courses and results from any university in the world, are securely stored and universities can exchange with each other what you have done, fully automatically? There would be no need to get transcripts printed, mailed, certified etc anymore. It would be done entirely online securely.

Let’s build such an imaginary system using ASP.NET MVC and WebAPI, and expose that over an ESB like WSO2 ESB. You don’t have to use any ESB. You can just run the ASP.NET app directly. ESB is there to offer secure, reliable exposure of the services.


Let’s define what features we want on this system: (SDB – an imaginary org) is an organization that offers secure online storage and controlled exchange of student records between universities, colleges, schools and other types of Educational Institutes (EI). An EI can establish a secure interface with Student Data Bank and upload its Courses, Programs, Students and Student’s Course Results. Then another EI can request a Student’s Record to be fetched from the offering EI. Student Data Bank (SDB) takes care of securely validating the request and fetching Student Records from an EI’s systems via the secure interface that EI has already established with the SDB. Thus SDB works as a broker between multiple EIs, empowering each to share student records with each other in a secure and auditable manner.

An awarding EI can initiate the process of awarding a degree to a student and in the process automatically establish whether the awardee student has completed all the required courses from the awarding EI, as well as from other EIs that the student claims to have attended required courses from. Each EI has a way to define what requirements their courses meet, so that courses can be accepted from other EIs automatically while awarding a degree to a student.

The Web site looks like this:

Fetch Courses

Clicking on that Fetch Course triggers a workflow, which fetches the courses from the University, via the interface integration done with the University.

Sequence Diagram

Read the full CodeProject article how this is done:


Build, deploy, anonymize config, zip package, git commit, push from a single command

gitautomateWhile working on open source projects, you have to frequently build your code, clean up all temporary files, remove your own passwords, connections strings from web.config, then create a binary deployment package in a zip format and then commit and git push to GitHub. Let’s automate all these using a configurable powershell script.

Here’s the full script.

First step, let’s define the parameters for the script with some default values:

param (
    [string]$solution = "OracleDashboard.sln",
    [string]$zipname = "",
    [string]$compressor = "c:\Program Files\7-Zip\7z.exe",
    [string]$folder = "OracleDashboard",
    [string]$deployPath = "..\Binary",
    [string]$commitFrom = "..",

Some description of these parameters:

  • $solution = path of the solution file relative to the script location
  • $zipname = name of the zip file.
  • $compressor = 7-zip’s 7z.exe file path.
  • $folder = the folder that contains the code, which is zipped.
  • $deployPath = relative path where the zip file will be moved to.
  • $commitFrom = relative path where the script will run git commit and git push from.
  • $comment = A comment for the git commit.

The first thing the script does is look for the solution open in Visual Studio and close it. You can comment this section out if you want. But if Visual Studio is open, then /obj folder cannot be deleted.

# If visual studio has the solution open, close VS, as we can't delete obj folder while it is open
$windowTitle = $solution.Replace(".sln", "")
$vsProcess = get-process | where {$_.mainwindowtitle -match $windowTitle -and $_.ProcessName -eq "devenv"} 
if ($vsProcess.Length -gt 0) {
    Write-Host "Visual Studio has this solution open. Closing..."
    $vsProcess | ForEach-Object { $_.CloseMainWindow(); }
    Sleep 5
    Read-Host "Press ENTER to proceed if Visual Studio is closed"
    $vsProcess = get-process | where {$_.mainwindowtitle -match $windowTitle -and $_.ProcessName -eq "devenv"} 
    if ($vsProcess.Length -gt 0) {
        Write-Host "Visual Studio still has the solution open. Aborting."

Next step is to do some spring cleaning:


if (Test-Path $zipname) { rm $zipname; }

# Clean up deploy folder 
rm $deployPath\*.* -Force -Recurse

First remember the current path. We have to come back to this path after we are done. Then remove the zip file if it already exists. Then cleanup the $deployPath. You can remove this if you want to keep old deployment packages. Then you have to handle generation of unique file names for the packages.

Now, let’s build and remove /obj folder:

# Build new version
msbuild /verbosity:minimal $solution

# Delete obj
if (Test-Path $folder\obj) { rm $folder\obj -Force -Recurse }

Next step: remove all sensitive information from the web.config, which includes connection strings, authorization block, appSettings entries etc. This is all up to you to customize:

# backup the web.config and remove sensitive entries before pushing to git, eg connectionString
[string]$filename = gi $folder\web.config 
[string]$backup = [System.IO.File]::ReadAllText($filename)
$xml =
$backup $xml.PreserveWhitespace = $true foreach($n in $xml.configuration.connectionStrings.add) { $n.ParentNode.RemoveChild($n); } # Anonymize any sensitive appSettings entry foreach($n in $xml.configuration.appSettings.add) { switch($n.key) { "Password" { $n.value = "Password" } } } # Remove authorization blocks $xml.configuration.'system.web'.authorization.RemoveAll() $xml.Save($filename)

Finally, let’s run some regular expression check to ensure the web.config does not have any sensitive information left accidentally. Again, this is all up to you to customize.

# verify if web.config still contains any sensitive info
[string]$config = gc $folder\web.config
if ( ($config -match 'connectionString="\w+') -or ($config -match 'users="\w+') ) {
    Write-Host "Configuration file is not cleaned."
    # Restore web.config
    [System.IO.File]::WriteAllText($filename, $backup)

Now time to compress the source folder and create a zip file using 7-zip.

# Compress the solution folder and copy to deploy folder
cmd /c $compressor a -tzip $zipname $folder -r 
cmd /c copy $zipname $deployPath /Y
cmd /c del $zipname

Finally git commit and push:

# Commit and push to GitHub
cd $commitFrom
git pull
git add -A *.*
git commit -a -m $comment
git push 

And last step is to restore your own web.config, that was anonymized:

# Restore web.config
[System.IO.File]::WriteAllText($filename, $backup)

That’s it. Now all you have to do is, just hit ./gitpush.ps1 from Powershell command line and you are done!

Real-time Oracle Database Monitoring Dashboard in ASP.NET

Oracle Performance Dashboard (OPD) is a small ASP.NET website that shows you performance & problems of one or more Oracle instances in near real-time. It uses the Dynamic Performance Views (DPV) and runs some popular DBA scripts in order to get meaningful, easy to understand information out of the server. You can use it to quickly spot blocking queries, who is blocking who, expensive query that are consuming high CPU or disk, see if there’s unusual locks, very high disk activity and so on.

Dashboard - Full


You can see a live demo of this from here:

Get the code

The binaries are here, which you can just extract into an IIS folder, put the connection strings in the web.config file, and you are ready to roll. No need to install any Oracle client software on the server.
GitHub Project Binaries

You can get the source code from the GitHub project site:

Feature walkthrough

OPD comes with the following features in V1:

  • Summary of all your instances in one screen, showing important indicators on each instance. Quick way to check if all your databases are doing fine or not.
  • Instance Dashboard showing details of an instance:
    • CPU usage on the OS.
    • CPU consumed by each session
    • Important System Statistics like Buffer Gets, Parse to execute ratio which would indicate some common problems on the server.
    • Sessions and what the sessions are doing, how much resource they are consuming, whether they are hogging the disk or not.
    • Waits, Blocks, Locks, deadlocks that make database suffer.
    • Historical analysis on the databse showing you some very useful stuff:
      • Most expensive queries in terms of CPU and Buffer Get, which are immediate convern for your devs to sort out.
      • IO usage on data files. You can see if some data file is unusually loaded and getting hammered by physical IO.
      • Tablespace usage. Alerts you if some tablespace is over 85% full.
      • Stale stats on tables. You should always keep this clean.
      • Killer Indexes that will actually blow up your system and confuse Oracle query optimizer. You need to get rid of those indexes and rewrite queries that you thought will hit those indexes for better performance. They won’t. They will kill your database.

Web-based real-time SQL Server Performance Dashboard

SQL Server Performance Dashboard (SSPD) is a small open source web app that shows you performance & problems of one or more SQL Server instances and their databases in near real time. It uses the Dynamic Management Views (DMV) to gather useful data from the verbose output and combines them with utility stored procs in order to get meaningful, easy to understand information out of them. You can use it to quickly spot blocking queries, who is blocking who, expensive query that are consuming high CPU or disk, see if there’s unusual locks, very high disk activity and so on.

See a live demo:

You can read details about the tool from this CodeProject article:

The binaries are here, which you can just extract into a IIS folder, put the connection strings in the web.config file and you are ready to roll:

SqlServerPerformanceDashboard GitHub Project Binaries

Or you can get the source code from the GitHub project site:

Utility to make important windows remain always on top

Do you sometimes fail to notice Outlook reminder window? Do you wish a chat window would remain always on top of other windows so that you never miss a message? Do you want to have a notepad always on top so that you can take quick notes anytime, while working on other apps?

We have an app for that.

It runs quietly on system tray and monitors open windows. Whenever it finds a window that you want to make always on top, it does that:

You can use this AlwaysOnTop utility to make any window remain always on top of other windows.

Get the tool and details here:


Sit back and relax, let Sharepoint remind and chase your team

Sharepoint Task List is a great place to record tasks for your team members. However, once you have recorded and assigned the tasks, then the fun begins. You have to remind your team members repeatedly about what they need to do today, what tasks are overdue, what’s coming this week and so on. Here’s an app that takes that chore away, and you can sit back and relax, while it will reminds your team members as many times as you want. Your team members will get an email like this, reminding them of their assigned tasks:

The email templates are fully customizable. You can define exactly how the email should look like. You can also include custom fields from sharepoint taks items to appear on the email.

Read the CodeProject article here for details: