Our new Sandybridge dedicated servers come with SATA3 which helps maximize the 330 series drive performance.  Data sheets show that over SATA 3 (at 6Gb/s) these drives can achieve 450+ MB/s, and ~25k IOPS.

We now offer the larger capacity 120GB and 180GB Intel 330 SSD (Sandforce based) options replacing both the original Intel X25-M 80gb option.

The SSDs are ideal for customers with performance critical IO loads.  Primarily: databases on sites with many concurrent users.  The low latency and very high throughput reduce overall server load.   And let you maximize other resources besides IO (like CPU and bandwidth).  This has the cascade effect of allowing your services to do more work on the same hardware.   Combine with plenty of RAM and our new Sandybridge CPUs and the  SSDs are hard to beat.

The SSDs are available for order on any of our Dallas-based dedicated servers.  Select the model that best suits you at http://rimuhosting.com/order/startorder1.jsp?hom=t-ded-dal

First, order your second VPS. This will work between data centers, around the world, or even within the same data center no problems at all. Make sure the disk size is the same size on both VPS. Install the new VPS with a copy of the old one, you can do this in your rimuhosting panel. This step is not overly necessary, but may save you setup/config and initial data transfer.

MySQL Replication

First you need to edit your my.cnf , this will be located in /etc/ or /etc/mysql/ . Add in the following items onto both servers
master-host = 192.168.16.4
master-user = replication
master-password = slave321
master-port = 3306
master-connect-retry = 60

Make sure you change the IP to the IP of the OTHER server, so server 1 has server 2 IP and vice versa, it just needs to know what to connect too and how. Check that you replace the 'replication' with user of choice, and passwords of choice

Next you need to put the following into one server to give it an id
server-id = 1
replicate-same-server-id = 0
auto-increment-increment = 2
auto-increment-offset = 1

And this into the other
server-id = 2
replicate-same-server-id = 0
auto-increment-increment = 2
auto-increment-offset = 2

This identifies the servers with an id, and also sets auto_increment to be offset so you do not get clashes with trying to overwrite any data.
If you only want to replicate the one database you can add a line like this on both
replicate-do-db = databasename

If you want to sync the entire database then you leave that out. If you want to ignore syncing for just one database then do something like this
binlog-ignore-db=mysql # input the database that should be ignored for replication

Okay, next you need to check the binlog line is uncommented if it is commented out, or add one
log_bin = /var/log/mysql/mysql-bin.log

Remove any instances of skip-networking and if you have bind-address make sure it looks like
bind-address = 0.0.0.0

Save and exit the mysql config now, and login to mysql on the command line. You will need to add the user you just setup on those configs that will be used for replication. Run this on both mysql servers, changing the IP/user/pass for the ones in your configs.
grant replication slave on *.* to 'replication'@192.168.16.5 identified by 'slave';
Make sure the user for each server has the hostname of the other server they are getting connections from on both of those.
Now the hard work is done, restart mysql on both servers. Now all you need to do is login via command line and start each as a slave

mysql> start slave;
mysql> show slave status\G;

Slave_IO_State: Waiting for master to send event
Master_Host: 192.168.16.4
Master_User: replica
Master_Port: 3306
Connect_Retry: 60
Master_Log_File: MASTERMYSQL01-bin.000009
Read_Master_Log_Pos: 4
Relay_Log_File: MASTERMYSQL02-relay-bin.000015
Relay_Log_Pos: 3630
Relay_Master_Log_File: MASTERMYSQL01-bin.000009
Slave_IO_Running: Yes
Slave_SQL_Running: Yes
Replicate_Do_DB:
Replicate_Ignore_DB:
Replicate_Do_Table:
Replicate_Ignore_Table:
Replicate_Wild_Do_Table:
Replicate_Wild_Ignore_Table:
Last_Errno: 0
Last_Error:
Skip_Counter: 0
Exec_Master_Log_Pos: 4
Relay_Log_Space: 3630
Until_Condition: None
Until_Log_File:
Until_Log_Pos: 0
Master_SSL_Allowed: No
Master_SSL_CA_File:
Master_SSL_CA_Path:
Master_SSL_Cert:
Master_SSL_Cipher:
Master_SSL_Key:
Seconds_Behind_Master: 1519187
1 row in set (0.00 sec)

If there are any errors they should show up here, and if you get stuck at all let us know, we are happy to help out!

File Replication

For file replication we are going to use Unison, its easy to install and use . Make sure you have passwordless ssh setup on both servers so they can connect to each other without passwords or prompts (preferably as root).

Now create a directory in /root/.unison/
This is where you will store your configs. You can have multiple configs and include them from each other if needed, but in this tutorial we will be only using the one for now called default.prf
Create a file called /root/.unison/default.prf and insert something similar to the following

log = true
logfile = /var/log/unison.log

auto=true
batch=true
confirmbigdel=true
fastcheck=true
group=true
owner=true
prefer=newer
silent=true


Now you should be able to run 'unison' from the command line and see any output, make sure things are ok. Pays to test a few times, I had errors regarding tmp files which was why i ignored those. No need to sync a cache for a website either if that is what you are syncing.
Despite this running on one server, it will sync both ways, so this only needs to be run on the one server not both (unless you have multiple servers).
When you have the unison as you want it to run, add that into a crontab and have it run every minute or two.
I have mine run only every 10 minutes or so, but you can have it running every minute if you need, or even have it triggered by another method if needed.
Here is my crontab entry

*/10 * * * * /usr/bin/unison > /dev/null 2>&1


Pays to create and delete files to test this on both sides to make sure it is working also.

Now you have 2 servers replicating each other. This means you can now add DNS so that doing a DNS lookup on a website shows 2 IPs, this means people will be spread over the 2 servers instead of the one, lowering the load on a single server. You can scale this up to multiple servers if required, and then take servers out of the array if they need work, without having downtime.

Things to note:
You may also want to sync configs (eg apache configs, websites, passwd files for users, etc).
Master/Master replication can be buggy - especially if you have very busy servers. There may be clashes on updates on both servers at the same time or things getting out of sync. This is why monitoring is critical to make sure it is going ok. For some sites it may be more suitable to have a single mysql on its own for some sites, and give it huge memory and CPU instead.
Other alternative ways for replication are described here http://dev.mysql.com/doc/refman/5.1/en/mysql-cluster-replication-multi-master.html
As a general rule, there is no one fix or solution for scaling upwards, this post was just to get you started on some of the basics.

If you need any help with this, just pop in a ticket and let us know what we can do for you.

This new platform will enhance any large scale dedicated or virtualized deployments with access to 24 hyperthreaded cores and support for 128GB RAM or more in a 1U form.

Anandtech have shown about 40% speed boosts over the previous generation E5560 processor for common server loads, up to 70% increases for some computationally intensive tasks.  They say:

Our conclusion about the Xeon E5-2690 2.9 GHz is short and simple: it is the fastest server CPU you can get in a reasonably priced server and it blows the competition and the previous Xeon generation away.

The servers are available for ordering in Dallas at http://rimuhosting.com/order/startorder1.jsp?hom=t-ded-dal in the dual CPU section under "2 x 2012 Sandy Bridge Xeon"

You can configure the server with up to 96GB of memory.  There are dual power supply and redundant power options.  Plus our standard option of private networking.

We try to keep our customers informed. One example is a recent notification from Parallels affecting some older versions of Plesk.

However it is a great idea to keep track yourself. We recommend signing up to the relevant announcement channels. Some handy examples are below.

Plesk email anouncements or rss feed

Virtualmin Profesional issue tracker

Webmin email anouncements or Webmin/Usermin community support

If you have questions at any time about enhancing the security of anything to do with your server please do get in touch with us

The people there were great, and i ended up meeting and talking to many fascinating people about all sorts of code, servers, set ups among other things.
I was a little nervous that i only had 30 minutes to fit in a lot of information, and i had to dumb it down a fair bit for those with less technical knowledge to make it understandable.

I did manage to fit the talk in with no problems at all, it only took 45 mins, plus question time! Nobody seemed to notice the missing 15 minutes from their clocks luckily, and everyone appeared to understand and have great questions about how it worked.

Because scaling is really specific, the hardest part i came up against was really trying to describe something, and explain something, in a way that was non-specific. Anyone who knows servers, knows that there is no one solution to suit everyone. How ever there are a few common areas you can usually look into.

For WordPress one of the bet things you can do is deactivate and DELETE any unused plugins or themes. Get a developer to check them, a broken one can take down your entire server easily! Not deleting them leaves a potential security hole should that plugin be discontinued, those files are still accessible via the web remember!

A few things identified were setting up things like a Cache (or Cache Plugin), enabling the wordpress object cache in wp-config, and moving as many things offsite as possible.
By moving things offsite, i mean things like video content, photos, music etc.
Why spend hours of your time setting up a shockwave/html5 video player, converters, and using disk space to play video when you can just use youtube? This is what they do for a living, they have extensive programmers who specialize in this, so it stands to reason they are very good at it, and it will save you a lot of money and time. The same method works for Photography, Music, or any other large files if you can do it. It will save you a lot of pain and money later on down the track.

Another key thing for scaling a single server larger is setting up HyperDB, then having 2 separate databases on other servers. This offloads database offsite, but without the difficulty of having to setup MySQL replication.

If you are using WordPress and want to know more, check here for your nearest WordCamp http://central.wordcamp.org/

PV-Grub lets you run a custom kernel on your server.  Rimuhosting is all about flexibility and PV-Grub is one feature that will give our customers a lot of different options.  Custom modules, latest kernel, old kernel, kernel development, lots of good stuff.

We continue to offer our own kernels.  These run external to the VPS. In these we enable most common options (but occasionally a customer would require an option we had not enabled). And in that case, if the customer lets us know we would often just go build a custom kernel for that customer with the options enabled.

PV-Grub is an option for you if you need to run some particular version of some kernel or module for a particular reason.  Most customers will not need to touch their kernel setup (the externally provided kernels work really well for a lot of people).

With PV-Grub, you are in control of the kernel you use, and can keep that updated as you see fit and load whichever modules you would like to run.

With our provided kernels then we get to do a bit of testing.  To ensure that the servers operate reliably on that particular kernel.  And to enable to us automatically upgrade kernels over time.  For people that do not want to fuss about doing that even if there are security exploits discovered.

To switch to PV-Grub or to find out if it is right for you check:

http://rimuhosting.com/knowledgebase/rimuhosting/pv-grub

For example, you will need to install grub and a kernel.

I have now added putting the site into maintenance mode whilst upgrading. This just involves adding a .maintenance in the document root.

It pays to check each site after the upgrade to make sure plugins all worked (and upgrade plugins before the upgrade works also!)

Let me know if you have any bugs at all, or any problems.

Here’s your script to upgrade them all.

wget http://b.ri.mu/files/wordpress-upgrade.sh
bash wordpress-upgrade.sh

MD5
07fb381feaf0d57daeb2b73017d0057d wordpress-upgrade-3.3.2.sh

SHA1
c5318dd294626ccac3873f62c0e03c3882c27d1d wordpress-upgrade-3.3.2.sh

If you find any bugs or problems, just let me know at liz at rimuhosting dot com. I have now renamed the script to wordpress-upgrade.sh for easy future memorability (using symlinks to link to the latest version)

For a limited time and while stocks last we are offering a $100/m discount on these servers.

• 2 x Intel Xeon Quad Core 2006 Clovertown or Harpertown CPU 1.6GHz+ (specific model would depend on available stock)
• Can take 4-24GB of memory.
• RAID as standard.  The servers can take up to 4 x 2TB disks.
• We can set it up as a regular dedicated server, or as a virtualization host running one or more virtual servers.
• Pricing from $199/m (including the discount)

Just mention this blog post in the order comments when you order.  You can order at http://rimuhosting.com/order/serverdetails.jsp?plan=DSRHTXXF1

Here is a picture of a few of the office staff (not all of them) located in Cambridge wearing them.

Contact us at support to arrange the exchange!  To get our T-shirt you can order directly at http://www.cafepress.com/rimuhosting and we can give you a hosting credit.  Cafe Press is good because you can order the size and colors and designs you prefer easily.  If you are in NZ then we can post you one from our stock at the office.