Andrei Rînea's questions

I would like to host a single physical machine two copies of the same very site to minimize downtime during upgrades.

I think I could put a load balancer in front of the machine, assign two IPs to this machine and set the IPs in the load balancer.

The issue is when I need to update the site (its code not the infrastructure). I think I could minimize downtime by taking out one of the site's 'instance', upgrade it then bring it back online and do the same with the other one.

Is this feasible?

Is this worth it?

I am aware that this question might be closed as "not a real question", "subjective" or even "argumentative". If this is a mistake, I am sorry.

I am not good at all at networking but I'm trying to get better and learn more. (you might know the saying : "the first step to intelligence is acknowledging that you are stupid" :P).

Network-wise what happens after I type www.google.com in my browser and press enter?

Here's what I know so far :

• The browser application makes a DNS request to find the IP of www.google.com. It might find 209.85.148.105 (that's what has just been resolved for me now).
• Then it connects via TCP/IP to that host on port 80 (default port for HTTP) and sends an HTTP request like GET /
• Receives HTML data and displays it (along with loading additional resources such as .css, images files etc.)

What I want to find out is how it connects, step by step, to a Google server. After the request goes out of my computer and my ISP it finally hits google hardware at some point.

• Is there a load balancer?
• How many of them?
• Are there multiple levels of load balancers? i.e.: a load balancer of load balancers of load balancers...
• How powerful is an end web server? Could it be a lousy 500 MHz CPU machine with 512MB RAM and 10 GB HDD? Or maybe an 8-core, 16GB, RAID-0 SSD beast?
• What kind of redundancy do they use at network level?
• What happens if load balancer(s) fail?
• What does it take to make www.google.com unavailable? (No, I don't want to sabotage or sth..)

I am curios whether having two or more of these instances on the same machine help using more RAM and more (logical) CPUs.

By default an instance of SQL Server Express would only use 1 GB of RAM and 1 cpu only.

Having a machine with 4 cores and 4GBs of RAM, could 4 instances be better? Or the Express edition would detect this and enforce that all instances would use 1 GB of RAM (summed) and only one core?

I also hear that you can't install more than 16 instances on a machine.. Could this be true?

I am trying to design a web site with backup in mind. Large media elements would be stored on the file system and a daily backup would be done both for the DB and the file-system (which would hold the media elements).

While backing up the DB seems straight-forward I am curious about backing up the file system. WinRAR allows incremental backup and it's nice and dandy but I wish the site would be fully-usable while the file-system backup is taking place.

What if this (FS) backup takes 10 minutes and a user adds/modifies/deletes a (dozen of) file(s)?

SQL Server 2008 R2 guarantees the consistency of a backup by having a two step process.

1. Backups all the data up to the backup start point
2. Backs up the rest of the changes since the start of the backup and it's end

I don't know if something equally awesome could be done to the file system...

I am expecting to have a long backup on a certain database and I was wondering if the application can access (read only? read/write?) it while it is being backed up.

If read/write access would be available then how will consistency be achieved? Probably only a snapshot of what was available at the time of the start of the backup will be included in the backup?

Or will the database be non-operational during backup?

I have a VPS with PowerShell installed. How do I know if it's a RTM or CTP?

I ran $host.Version and I got :

Major  Minor  Build  Revision
-----  -----  -----  --------
2      0      -1     -1

WTF are the minus-one's? I've never seen negative numbers for any of a version fragments..

I rented a VPS / virtual machine with Windows Server 2008 with 16 GB of HDD.

The Windows install was fresh so I went on to install the automatic updates, patches etc. I soon found out that I now have 200 MB free as SP2 couldn't install reporting free space too low.

I've uninstalled all features and roles and every program and I am now at 400 MB free.

How can I go about finding out WHO is occupying my space and HOW / IF I can fix this?

Thanks :(

Later Edit 1 : Initially the hoster put up a Windows 2008 R2 VPS image which had MUCH more available space (I think 7 GB of the same 16 GB)... could it be that R2 uses much less space?

I found in a features table for SQL Server 2005 a row named "Star query optimization" but it is not explained anywhere.

Could this be a feature that optimizes a query like :

SELECT * FROM Table

On a Windows 2003 Server I have an application that processes requests and stores them in a folder as a queue where a second process comes and processes these stored requests. However at times the number of files tend to reach 100,000.

Using Windows Explorer or even dir at the command prompt doesn't seem to be much of a solution for me to find out how many files there are stored. It also takes too much time, CPU and I/O.

Any suggestions?

LATER EDIT : Could it be that IIS is not case sensitive for URLs and Apache is and someone could use this information to further increase the confidence factor in OS/WebServer detection?

I was reading about the useful UrlScan tool here and came across

IIS 6.0 does not include the RemoveServerHeader feature because this feature offers no real security benefit. Most server attacks are not operating system?specific. Also, it is possible to detect the identity of a server and information about the operating system by mechanisms that do not depend on the server header.

Of course this stirred my curiosity since I don't understand how you can detect the OS and/or the web server without using the server header. However extensions (.php, .asp, .aspx, .do, .py etc. etc.) cannot be the answer to this question nor looking for "__VIEWSTATE" or similar hidden input fields in the response content.

Is there some secret way that I don't really know about?

EDIT 1 : I assume customized error pages (not the default ones which clearly show the web server; knowing the web server gives you a strong clue about the OS too)