The Zeus Blog: Application Delivery, Cloud Computing, Load Balancing, Virtualization

Role reversal in Cloud Computing part 2

Previously, we considered how Application Developers may be the point of entry for cloud computing into your organization.  In this posting, I'll consider the effect that might have.

Cloud Computing is not just about cost savings

The role of the applications team changes when they start to use cloud platforms.  Now, they control not just the application code, but large portions of the application infrastructure it runs on too.  Apps guys are smart, and basic infrastructure isn’t hard.  Your good apps guys probably run their own infrastructure from home or from some service provider already, and know enought about TCP and HTTP and Firewalling and to hold their own with their network colleagues.

Perhaps this sounds like a recipe for disaster, but with sufficient support and buy-in, you can make this shift in responsibilities work.  Finally the applications teams can drive the infrastructure, leaving the cloud provider to make the hard hardware decisions.

In fact, it’s a potentially a catalyst for great success.  Take load balancing as a case in point.

Load Balancers

Modern load balancers (a.k.a. “Application Delivery Controllers”) are immensely capable devices that can manage and transform network traffic in a multitude of ways to address all sorts of application problems.  However, there’s a problem.  They are often in the wrong place and they can’t be moved.

The majority of load balancers are hardware appliances.  They sit in the datacenter, owned by the network team, with a massive organizational firewall around them.  Their scripting and traffic management capabilities are blunted because they are in the wrong hands.

Suppose that an application developer has a problem he needs to solve.  Perhaps there’s a bug in an application that is provoked by particular requests and he needs to seal it off pending future investigation.  Perhaps there’s a clustering problem he could bypass with appropriate session persistence in the load balancer.  Maybe he has to support some new client software that uses a slightly different protocol variation that requires either an application modification or a simple rewrite rule in the load balancer.

In each case, the problem can be solved using the load balancer more quickly and effectively than by modifying the application code, but this assumes two things.  Firstly, the developer needs access to a load balancer to develop and test the fix, and secondly he has to negotiate his way through the organizational firewalls to place his change request and wait for it to be approved and applied.

Imagine instead that your load balancer / application delivery controller resided in the cloud as a software service.  Your application developers could get unrestricted access to a dedicated load balancer (on a pay-by-use or zero-cost basis) to develop solutions and use the functionality more effectively in his applications.  This is entirely possible if you choose a cloud provider who offers this advanced load balancing functionality as service (e.g. Joyent), or if you use a software load balancer rather than a hardware one (and run it in your cloud of choice), particularly one with free licenses for your application developers.

Make your application infrastructure ‘cloud ready’

This ‘empowering’ of your applications team for the benefit of your business is not a consequence of cloud computing.  It’s a consequence of making your application infrastructure ‘cloud ready’, whether or not it ever gets deployed to a cloud. 

Making your infrastructure ‘cloud ready’ entails a subtle, discreet shift in power to your applications team.  Audit your application infrastructure, picking out any components of your composite application that are implemented in hardware appliances – firewalling, intrusion detection, web application firewalls, load balancers etc – and replace them with software alternatives when the next technology refresh comes round.  If necessary, walk round your datacenter with a pack of post-it notes tagging any brightly-colored single-purpose appliances with the label ‘anchor’, because those appliances are going to weigh your application down and prevent you from ever moving it to the cloud.

When you do come to deploy software alternatives to your legacy hardware appliances, you will naturally deploy them on the same type of server hardware that you use for your application servers and application code.  So it’s a natural move to hand responsibility over to your applications teams to manage them from then on.

By doing so, you achieve two things.

You make your infrastructure ‘cloud ready’. Next year, when your CFO reads the article in his airline magazine explaining how MultiCorp saved $millions and delivered new services faster than their competitors as a result of moving to a cloud architecture, and then asks you “why can’t we do the same?” you’re lined up to be your company’s next hero and the CFO’s new best friend.

A ‘cloud ready’ infrastructure sidesteps the ‘breaking down silos’ challenge that many organizations struggle with.  It goes directly to the heart of a key challenge, empowering your most innovative, valuable employees to create more new services, more quickly for your business. 

Like so many advances in the IT industry, Cloud Computing is much more than a technical solution to a technical problem.  Like client-server, and SOA and Virtualization before it, cloud computing can be a catalyst (or just an excuse) to shed legacy shackles that hold your business back and effect real change for the better.

Owen Garrett

Zeus Podcast - How can you make your web applications more secure from malicious attacks

Welcome to the first in a series of podcasts run by the team here at Zeus. This week's edition focuses on the recently launched Zeus Application Firewall Module (ZXTM AFM), with an introduction to the solution and how it can be deployed in conjunction with ZXTM. You can listen to the podcast in .mp3 format by clicking on the link below.

Hosts: Mark Gyles, Owen Garrett

Download: Zeus Podcast ZXTM Application Firewall Module

Why do you need a web application firewall?

Many web-based applications are vulnerable to a range of attacks that cannot be protected against by traditional network firewalls. For example, cross site scripting, cookie tampering, forceful browsing and SQL injection attacks can be used to steal information and disrupt user's transactions, but because these attacks are hidden deep within the HTTP or HTTPS transactions, they are very difficult to detect and prevent.

Zeus has extended its offering by providing ZXTM AFM (ZXTM Application Firewall Module) which ensures business-critical applications are secure from the threat of malicious attacks and to protect online customers personal and financial information.

The ZXTM AFM solution is designed specifically to help organizations comply with the Payment Card Industry Data Security Standard (PCI DSS). This is a global requirement that organizations who trade online must meet, regulating the processing, transmission and storage of credit card information.

ZXTM AFM helps organizations meet these requirements and can be easily configured with additional security policies to detect and prevent future attacks to your applications.

We would like to hear from listeners who may have topic suggestions that they would like to see offered in the podcast format. Please email: blog(at)zeus.com with your suggestions or any other feedback you may have once having listened to this podcast.

Mark Gyles

Role reversal with cloud computing

This is part 1 of a two part blog post based on a presentation given yesterday at CloudCamp Paris.

Some unexpected benefits of cloud computing

Gartner calculated that medium to large enterprises could spend as little as 20% of their annual IT budget on ‘value creation’ – making the new stuff (applications and services) that differentiated the business from its competition.  Up to 80% of the annual IT budget went towards “keeping the lights on” – the capital and operational costs of the infrastructure to host the applications.

Hand in hand with budget is power and influence.   These figures betray the deep-seated and unhealthy imbalance in power in many large organizations which are held to ransom by costly, inflexible business infrastructure that is very resistant to change.

If you work for a smaller organization, these figures may leave you in disbelief, but the infrastructure treadmill is almost inevitable as your organization grows.  It takes some very smart thinking (the sort of thinking that lead Amazon to create EC2) to break away from this.

One of the key aspirations of Cloud Computing is that it will allow organizations to exchange some of those enormous capital infrastructure costs for pay-by-use operational costs, and through elasticity and improved infrastructure utilization, reduce your overall costs.

However, just as virtualization brings more benefits than just server consolidation, so the Cloud Computing approach can bring more to the table than just pay-by-use and elasticity.

Who adopts cloud computing?

In an organization, you might think that the infrastructure and operations teams are the first to embrace cloud computing, but in fact, they are usually so far out of the room that you’ll find them standing on the edge of the window sill with their fingers in their ears singing nonsense verse very loudly.  Cloud Computing threatens these teams with serious loss of control, power and influence, and they will spread as much FUD as they can to keep it off the agenda.

Instead, the first adopter and most vocal proponent of Cloud Computing is often the Application Development team.  Cloud computing lets them bypass the organizational barriers and obstacles that the datacenter guys put in their place.  Watch how the balance of power in your organization begins to shift.

Picture the scene...  the apps team need 8 servers for a 15 day’s worth of QA testing before releasing a new service.  They approach the business manager who commissioned the service:

“The datacenter guys have told us there’s a lead time of 2 months to acquire and rack up the servers we need (yeah, sorry, we forgot to ask two months ago), and they said some sniffy nonsense about the datacenter being almost ‘full’.  I’ve been using a cloud provider for some other stuff, and I can get the 8 servers we need from them in 10 minutes.”

Now the business manager is sold on Cloud Computing.  He approaches his CFO:

“Your datacenter team has quoted me $10,000 for the 8 servers I need for QA testing.  My apps guys can get them from their cloud provider for $40/day.  We only need them running for 8 hours a day, so it’ll cost about $200.”

The CFO has to choose between putting $6,000 of depreciating assets on the balance book and accounting for $4,000 of internal costs, and signing off on an extra $200 on someone’s expense report.

Granted, this is just a QA system where the benefits of cloud computing are particularly clear, but you can imagine the conversations about the production systems that follow over the next few months.

So that's the potential cost benefits... what else?

Next week I'll disuss what might happen when your apps guys get free reign on your application infrastructure...

Owen Garrett

Zeus at CloudCamp Paris, June 11th

This coming Thursday sees Paris play host to the next in the series of CloudCamps.

For those of you already booked to go and those considering a trip, you may be interested to know that Zeus' Owen Garrett will be presenting one of the Lightning Talks, "Load balancing in the Cloud". Owen finds himself in good company with talks from the likes of: Sun Microsystems, RightScale and Orange. The usual type of open discussion sessions will follow the Lightning Talks, topics include: What is Cloud Computing?, The Open Cloud and Clouds Security.

We hope to see you at the event. Owen will be on hand throughout the event should you have any questions about Zeus or our solution offerings.

Further details, registration and a full agenda can be found here.

Mark Gyles