Login's Today...

[Scott Greczkowski]

You might notice today that you might have had to log in, this is OK as we are currently changing our web serving technology. As we work you may be required to log in again.

Things will return to normal when we complete our work.

What are we doing?

We are adding a second web server to our setup which will improve page load times for all of our members! In addition we are adding load balancing to these web servers to speed things up even more!

This work is expect to last most of the day however we expect to be OPEN for the majority of it, to serve YOU!

 

[Scott Greczkowski]

Just FYI, there MIGHT be a slowdown when we mirror all the files from our current webserver onto the new webserver.

Although since the data transfer is happening on its own private network I would hope that the file copying would be invisible to everyone.

 

[PopcornNMore]

Thank you Scott for all the work you do to improve this great forum.

 

[dvrexpander]

Yeah good news.. it was loading so slow sometimes I felt like beating my monitor senseless!

 

[whitewolf8214]

ty scott keep up the great work

 

[wallyhts]

Just FYI, there MIGHT be a slowdown when we mirror all the files from our current webserver onto the new webserver.

Does that mean your running two web servers and a load balancer or retiring the old web server? I think your idea of running ads in iframes is fantastic.

 

[Scott Greczkowski]

Does that mean your running two web servers and a load balancer or retiring the old web server? I think your idea of running ads in iframes is fantastic.

We have two servers one currently does web, mail, name service, secure file transfers and a few other things.

Our other server currently just does one funcation and that is the database server. That server has 3TB of space with 2 Quad Core Intel XEON processors. The server load there is normally 0.10 or under. Most of the time it is around 0.03.

We are loading web serving software on the database server with special load balencing software to take advantage of the almost totally unused CPU power on the database box. This will speed things up nicely. It is expected in this configuration that the load on the database server will still not go above 1.0 so there is a lot of room for growth.

In addition we are loading the latest caching technology as well which will also help out greatly.

 

[Scott Greczkowski]

As you can tell we are still working...

Our work got hampered a bit as our ISP had an issue (second issue in two days) which made the servers unavailable.

Here is the note from the ISP explaining the outage.

COLO4 UNPLANNED NETWORK EVENT NOTIFICATION:
________________________________________________________________

EVENT ID: COLO402132011-911-A

DATE: 02/13/2011

EVENT START TIME: 12:00 PM CST

EVENT END TIME: 1:00 PM CST

SERVICES/EQUIPMENT: Colo4 Premium Network

TYPE OF EVENT: Network Hardware Performance Incident

IMPACT OF EVENT: Reduced throughput, increased latency, service interruption

ACTION TAKEN: EMERGENCY REBOOT OF BORDER ROUTERS, router reconfiguration.

RESOLVED?: YES


Event Description:

At approximately noon today, an FIB TCAM exhaustion event resulted in the need to perform an emergency rolling reboot of each of our Premium Network border routers. This resulted in a period of reduced network capacity and network interruption.

We have identified the underlying cause of this issue and corrected it by reallocating resources on the routers in question and modifying the configuration affecting the distribution of these resources.

Please let us know if you are experiencing any issues at this time. At this time, the issue is believed to be fully resolved.

This notification is being sent to ALL CUSTOMERS. This notification pertains only to the PREMIUM NETWORK. The Value Network product was NOT AFFECTED.

If you have any questions or concerns please address them to our NOC at support@colo4.com or give us a call at 214-630-3100, and reference event number COLO402132011-911-A.

 

[Lord_Vader]

Scott, What kind of web server do you run? IIS, apache, ..etc. Just asked because you may be able to turn on things like compression to speed up the site. ( seems fine to me already, but thought I would mention it. )

 

[Scott Greczkowski]

We were Apache 2.2, then yesterday we were Varnish with Apache which was setup more as a test to see how well Varnish worked... Varnished worked well but only people who saw improvement were those not logged in.

As of this morning we are now running Xgengx with Apache 2.2. We are still in single server mode as I type this as I sent our helper the wrong password by mistake to start the sync of files between the two servers.

Once the sync is setup and properly synced we have Xgenex and apache also setup on the other server we can then enable the load balancing software and begin serving web pages from both servers.

In addition I am going to be new CPU's for our main websrver which will take us from dual core AMD's running at 2.2 Ghz to dual core AMD 2.8 Ghz processors which should give us an added boost.

Why am I doing this? Because I want the best for you guys.

Thanks for your patience!

 

[Lord_Vader]

Do you use mod_gzip?

 

[Scott Greczkowski]

We use mod_deflate

WhatsMyIP.org HTTP Compression Test

http://www.satelliteguys.us/243330-logins-today-2.html is gzipped Original Size: 61.19 KB
Gzipped Size: 15.53 KB
Data Savings: 74.62%

 

[Lord_Vader]

Cool. Just thought I would ask.

Also, check your DB response times. Normally DB is a hog on resources. ( slow DB causes high CPU in the front end )

 

[Scott Greczkowski]

The database responsed time is not an issue. Normlly 0.01 or lower

We have a kick ass Database server.

I got to sat just the single server NGENX and Apache we have running at the moment is quite snappy. Hopefully it will be even more so once the other server is up and running.

 

[Scott Greczkowski]

How are things running now?

At the moment we are just on one server, but tomorrow at this time the second sever should be online and load balancing in place.

From here it appears the new NGENX web software is very snappy.

 

[yaz96]

Why am I doing this? Because I want the best for you guys.

AND you like new toys and to play with them!

 

[Scott Greczkowski]

Nope same old toys as before. We are just better using what we have. One server was sitting there not using 99.9% of the resources it has on it. Time to put that puppy to work.

As you can tell work is continuing today and it should also be done today besides some minor tweeking for best performance.

 

[jdspencer]

So once we see that logins are back to where they were, we will know that all changes have been completed?

 

[Scott Greczkowski]

We are working on the final changes now... (adding in the second web server.)

All should be normal in a little bit.

 

[AnthonyG]

From here it appears the new NGENX web software is very snappy.

NGINX..... (Engine-X)