#Computing on the cloud

I have recently been asked how to achieve load balancing on the IBM Smart Cloud as our customers endeavour to deploy more scale out web infrastructure.  To this end I have written this tutorial to demonstrate the relative ease with which this can be accomplished at its basic level. Load balancing on the IBM Smart Cloud is achieved by using IBM’s approved load balancing ISV image Riverbeds Stingray Product. I’m going to demonstrate by using the Stingray image on the Smart Cloud global catalogue how http load balancing can be achieved in under an hour ( the majority of which is waiting for the image to provision on the Smart Cloud).

In this tutorial I assume that you are confident in using the Smart Cloud portal to provision Linux instances and that you are also confident in managing them through a product like Putty using key based authentication if not please see a previous post Deploying an Instance on the IBM public cloud

Installing the Load Balancer on SUSE Linux.

So lets get started. Login to the smart cloud and deploy the Stingray Traffic Manager image of your choice. I am going to chose SUSE 64bit as this is the version I prefer using.


Once deployed wait for the image to become active.


Once the image is active, SSH into the image and perform the list of actions below:

  • Open ports required for the management of the load balancer $sudo /sbin/yast will open the systems management tool where you can open ports to enable web management. Port 9090 is required by the Stingray web management console. Port 80 is also required should you intend on load balancing HTTP web pages
  • Once this is complete, perform the following commands as root. To gain root use $sudo su root
  • Ensure you are the superuser.
  • Run the installation script: /home/idcuser/stingray-install
  • Run the configuration script: /usr/local/zeus/zxtm/configure
    1. Accept the license agreement
    2. When prompted for the license key enter: /tmp/stingray_devkey.txt
    3. Accept the default values for UNIX user and UNIX group
    4. When prompted with Would you like to restrict Stingray Traffic Manager management to one IP?, enter N
    5. When prompted for the cluster option enter C
    6. Enter a password for the Admin server when prompted. This password will be required to access the browser Administration interface.
    7. When prompted with Would you like Stingray Traffic Manager to start at boot time?, enter Y
    8. Verify the Traffic Manager starts (look for the message Starting Stingray Traffic Manager Software… OK)
  • Connect to the Admin server at the URL provided at the completion of the configuration script. Username is admin and the password is the one specified in step 6. Please see below web admin address in box.


Once you have the address of the web admin console use this to connect to the login screen, enter the password you created during the setup. Once logged in you will see the screen below.


Configuring Simple HTTP Load Balancing.

For this part of the tutorial I assume that you have something to load balance. For this I am simply using two apache web servers running on SUSE Linux. These have one web page each in htdocs with WebServer 1 and WebServer 2, to demonstrate that load balancing is occurring. I have found this a great method of testing load balancing is working as expected, if you want to use these files please feel free to download them from here: Webserver load balancing test index.htlm

To configure load balancing start by gathering together the DNS names or IP address of the webservers intended to load balance. From here click on Services and select Pool. This defines the load balanced resources (web servers) that the load balancer uses.


I have configured the pool to monitor the webservers using Simple HTTP. However, there are a number of other options available to you.  Once pool has been created add the second webserver and then click update.


Now Select Virtual Servers. We need to build a virtual web server which will be used as the point of entry to our load balanced web servers.

Start by giving the virtual server a name. Select the protocol which in this case is going to be HTTP, a port, which is 80, and then a pool which in this example is SmartCloud_LoadBalanced_Pool. Click create. Once this is created you can edit the load balancing settings. I have simply set this to round robin but you can setup load balancing based on link response time, weighted, predictive etc.


Go back to the home screen and click start in the services area that has appeared after setup. After a few seconds you will see the button you clicked go green and the Load balancer is up.


Now try accessing the web page you setup earlier, but use the load balancer DNS with HTTP. You should see the test webpage come up, refresh a couple more times and you will soon see the other web page demonstrating that requests are being sent to alternate web servers.


Hope this helps. We have really only scraped the surface of the product, so for more information check out the Riverbed website. Please feel free to leave comments.

Ever wanted your cloud environment to resemble your physical datacentre check out this little app built on Adobe AIR with plugins to Amazon’s AWS Console

This is a sub-set of virtualfront showing a couple of servers and my S3 bucket in one rack and my Elastic Map Reduce S3 buckets in another to give logical visual separation of the applications.

gregs toolkit

Take a look for yourself at http://www.gregstoolkit.com/

Also available for iPad

I’ve been looking for a while for a way of trying to provide analytical information on my various web sites and web portals. I looked at many in the space of a month and bumped into this in the last week been having a play with it. Integrated it into virtualfront.


From what i’ve seen so far I have been very impressed. Piwik has the capability to integrate into the web page. This then sends back information to a Piwik server either running as a separate server or on the same instance as your web server. The whole application is only 5.5Mb which is nice and light weight. It provides great multi tenancy for someone looking to use it with multiple websites and is very easily customisable to your company or brands. Check out Piwik’s live demo pointed at their forums at:



Web based login also customisable and you can setup user accounts based on the website being monitored. Also deployable as a bitnami custom image

Just been working with some Linux instances for a customer. Not new to Linux but found this simple little guide a great help to save the grey matter for other things.

One Page Linux Manual

Will the world survive a day without Wikipedia?


Ill tell you tomorrow…

I have now been working with cohesive FT VPN cubed for a number of months in both production and test / dev. The product brings a level of security to a cloud computing environment that allows for truly secure instance data transit.


The product is used to provide encrypted layer 2 network between different end points. This type of networking is now more commonly in cloud architecture as an overlay network. Cohesive’s overlay network product VPN cubed uses a VPN switch to connect secure tunnels based on certificates generated by the VPN cubed manger.


This product, and the support team behind it, I cannot speak highly enough of. This really does provide my customers with the four walls and locked door feeling of a datacentre. The product, when implemented correctly, can provide a high, and dare I say higher level of data transit protection than provided from a colocation rack in a datacentre.

The Cohesive FT VPN cubed image is available fully supported on the IBM cloud and also Amazon EC2, Elastic Hosts, Go Grid… you get the idea!

Just finished working on my first AQL Statements (Analytics Query Language). This is IBMs information extraction language for the IBM Big Insights Platform. This platform is an IBM wrapper to the open source hadoop map reduce solution. Just about to drop it onto my 20 node biginsights cluster! So what does this have to do with cloud. Workloads like this benefit from large quantities of processing power for short periods of time. To this end cloud computing can provide the perfect platform to provide cost effective hosting for cluster computing.

Some great resources

The first in a great series of tutorials


Big data University

The end game

Working today with a partner we ran into trouble while trying to workout the software licensing costs for Cognos. after a bit of digging found the below. Enjoy…

Taken from IBM Passport Advantage web site: http://www-01.ibm.com/software/lotus/passportadvantage/licensing_for_IBM_Cloud.html

The IBM SmartCloud Enterprise offering provides a hosted infrastructure service, including hardware, operating system, and services available on a pay as you go or reserved capacity basis. There are nine virtual machine (Instance) sizes available, and three licensing options for IBM customers to use IBM software. Customers can:

Access pre-configured virtual machine Images (Images) under the pay as you go (PAYG) option.
Access Images under the bring your own license (BYOL) option.
Bring and upload IBM software Programs under the bring your own software and license (BYOSL) option.

For more information about the IBM SmartCloud Enterprise service and the PAYG option, visit IBM SmartCloud Enterprise or email david.fearne@arrowecs.co.uk.

Customers can use software licenses, acquired under IBM’s International Passport Advantage or Passport Advantage Express Agreements (collectively PA or PA Program) with the IBM SmartCloud Enterprise service. For licensing requirements under either the BYOL or BYOSL use options, select the respective tab at the top of this webpage.

IBM Passport Advantage Customers can access or upload and install IBM Programs on Instances using Program licenses obtained from IBM. See the BYOL and BYOSL tabs for more information. For software Images and Programs which are licensed by Processor Value Unit (PVU), the Table of PVUs required per Instance below lists the number of PVU Entitlements required per software Image or Program for each IBM SmartCloud Enterprise (SCE) Instance size as of the Table publish date.


(a.) For more information about IBM Cloud Instances, visit IBM SmartCloud Enterprise

(b.) Certain restrictions may apply; consult the BYOL or BYOSL tabs for additional licensing requirements for software Images or Programs, respectively.

Note: PVU licensing requirements per Instance are subject to change without notice.

IBM Cloud is available through IBM.com/cloud. This is the landing page for all IBM’s cloud service offerings both public, private and virtual private. Today we are going to take a look at depolying an instance from the IBM Smart Cloud Enterprise which we will refer to as SCE. SCE is IBM’s IaaS offering and can be used to provision instances based on a pre-defined list of images. imageWe are Login to the SCE control panel by selecting Cloud portal login. If you don’t have an account send me an email at david.fearne@arrowecs.co.uk. The IBM cloud as I posted below is not an EC2 or GoGrid type of cloud the IBM cloud is managed and charged in a more traditional method using the IBM reseller channel. This has its pro’s and its cons as you might imagine but the main reason for this is that unlike the other cloud providers mentioned above, IBM has rules of operation that do not allow them to host certain business types such as gambling or pornography. This also enables the traditional IBM business partner value add model. The video below is a brief introduction to getting started with the IBM SCE Cloud.

Introduction to getting started with the IBM SCE Cloud.

This video has been cut down but total deployment length is 20mins. come back soon to see how we implement overlay networks to provide global private network, load balancing and high availability.

Getting up and running with a useful cloud service has never been so simple

Bitnami (odd name but great service) have gone to the trouble of creating a large number of pre-built either packages or entire images that can be deployed as or to common Linux operating systems or Windows. The packages include Ruby on Rails, LAMP and WAMP, WordPress, Joumla and many others. The very variety and speed at which you can deploy a fully working website, blog or application makes them very attractive to people looking to deploy web services quickly and with little or no further customisation. I have experimented with a few packages now and run the very blog you are reading this post from on a bitnami WordPress package on Linux. with a few teaks…

BitNami Stacks

BitNami Stacks make it incredibly easy to deploy your favorite open source software. BitNami Stacks include an open source application and all of the other software necessary to run it, such as Apache, MySQL, PHP or Ruby. All you need to do is select your deployment format of choice (Native Installer, Virtual Machine Image or Cloud Image) and either run the installation wizard or start the image. In just minutes, your new application will be ready to run. All Native Installers for the Stacks have been packaged using BitRock’s multiplatform installer.
There are other ways of using BitNami. If you choose to install a BitNami Stack natively on your system, you can download a base Stack and then use BitNami Modules for the added benefit of being able to install several applications on top of a single infrastructure stack. Learn more about BitNami modules. We also offer VMWare images and Cloud Images for most Linux-based stacks as a download option.

BitNami Application Stacks

SugarCRM Redmine Alfresco JasperServer
WordPress Drupal Joomla OSQA
Coppermine Photo Gallery DokuWiki Moodle eZ Publish
Gallery Mantis MediaWiki ocPortal