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

The Inevitable Shift Toward Virtualization and softADCs

Mike Vizard of CTOEdge recently wrote an article on the integration of virtual and physical switches, pointing out that the “process of managing all the traffic being generated by physical servers that can now support multiple virtual machines is getting untenable.”  We approached Mike to see if he was interested in learning how Zeus is helping companies to address that problem and to introduce him to our technology in general.  The interview (and resulting article “Breaking Down Data Center Boundaries”) brought up some good points that I wanted to elaborate on a bit more in this post.

Mike notes that the shift to virtualization presents tremendous opportunities for convergence in the data center.  He also notes the benefits of Zeus’ pure softADC approach saying, “What’s interesting about a software-only approach is that it, at least in theory, can give the people who manage virtual machines a lot more control over bandwidth, as opposed to having to ask dedicated network managers for assistance.  That capability essentially furthers the idea of convergence within the data center.”

While we certainly agree with Mike on those points, there are several additional compelling business benefits of leveraging a pure softADC – particularly with more and more companies moving to the cloud.  For one, a pure softADC enables companies to deploy as many virtual appliances as they want to, without having any physical boxes outside of that environment.  This in turn gives companies the advantage of scaling up and down as needed. 

Performance is another benefit.  Despite the myth that softADCs don’t match the performance of hardware solutions, we recently conducted tests of our Zeus Traffic Manager on Cisco’s UCS and achieved throughput levels of 18.2Gbps.  That’s faster than any hardware or software ADC solution on the market, including F5’s and Citrix’s.  (If you’re interested in seeing more specific test results, you can view our new whitepaper.)

This begs the question, why are some ADC providers continuing to release hardware-based solutions when market trends so clearly indicate a shift toward software-only platforms? 

Poorly performing websites can cost retailers more than just lost revenues.

Last week Hostway (one of our UK Service Provider License Agreement (SPLA) partners) released a note on the wires highlighting the results of research that they had commissioned with us, which was conducted by independent research company TNS.

Lost revenues

In a nutshell the research highlights that UK retailers are potential losing up to £5.65 billion per year in lost sales due to their websites performing poorly. According to the research, consumers are abandoning online transactions and the survey of some 2,033 consumers highlighted that on average they abandoned 5.5 online transactions per person over the last 12 months.

Much of this abandonment seems to be down to frustrations of respondents wasting time waiting for page or image loading to take place, compounded further by many retailers websites containing multimedia content.

Brand reputation and customer loyalty

An additional topic merits some discussion at this point. This is the impact a poor performing website can have on the retailers brand reputation and its customers loyalty. Shoppers, whether purchasing from the bricks and mortar stores or online will have expectations that are set by high profile retail brands. The last thing any retailer operating online needs is to fail to meet these expectations because of a poor performing website, let alone the thought that customers would leave the site to go to a competitor site. Shoppers have always been an impatient bunch. Last year, research conducted by Forrester Consulting and Akamai Technologies in the US concluded that the new threshold in terms of an average online shopper's expectation for a web page to load was two seconds, and then shoppers are off to the competitions’ site.

High profile website downtime issues

Unfortunately, a poor performing website is not the end of the story. In recent times some organizations (Harrods, Ryanair, Paypal and Barclaycard for example) have experienced some rather high profile website downtime issues damaging brand reputation and calling into question the importance some of those businesses have placed on the technologies used to support their online presence. Any planned or unplanned downtime will have a huge impact on an organization and its customers. With the rise in use of social networking sites such as Twitter a business should not think others will not find out about their misgivings.

Intelligently manage your web traffic

The good news is that there are flexible software technologies out there that can intelligently manage web traffic and can help to mitigate poor performing websites or unplanned downtime. Retailers - in-fact any organization - that operates online, can ensure the delivery of a robust and reliable service to its online customers without the worry of damaging brand, reputation and ultimately revenues.

'Web-stress'

It’s not just business that is adversely affected by poor performing websites. According to Neurological research provided by software firm CA and customer experience consultants Foviance, volunteers found the process of searching for products and paying for them particularly stressful. The research also proved that websites that were slow and difficult to use meant visitors had to increase concentration levels. Catriona Campbell, director at Foviance and a behavioural psychologist, said: "When online expectations aren't met, people quickly become agitated, confused and have to concentrate 50 per cent more than normal."

Worried that your poorly performing website is just losing you revenue...then think again. It might be making your customers sick too!

If you're a retailer and would like to find out more about how Zeus can help you deliver on your customer expectations and ensure your online presence is fast, secure and available. We are co-hosting a webcast on March 18th with Retail Week, Quocirca and 'I Want One of Those'. Full details and registration can be found here. 

Mark Gyles

CloudStorm London, UK, February 22nd

Zeus will be attending and presenting at the next event in London at Allen & Overy, One's Bishops Square. 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. Other organizations presenting include Rackspace and Quest Software's Joe Baguley, who will be presenting on the topic: Cloud - how are real organizations actually using it, and what do they want?

CloudStorm is a new style event set up by companies offering innovating solutions in cloud computing. At CloudStorm, you will receive in 100 minutes an update of the latest trends and products in Cloud Computing, followed by a mini expo and networking opportunity.

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

Mark Gyles

Bandwidth Management but not as you know it...Part 2

Previously, we discussed how many organizations view management of bandwidth from a network layer 4, WAN, network-admin, datacenter perspective and how they should start to consider how they might handle the issue from a layer 7 (application) perspective. I now plan to introduce the use of application delivery controller and specifically Zeus Traffic Manager as a way of managing this process.

Implementing and managing your bandwidth policies

As highlighted in the earlier post the reference article suggests a number of practical policies and actions to be taken to assist with freeing up bandwidth but does not expand on how this can actually be achieved and implemented. An application delivery controller is a great solution for creating these types of policies and implementing them.  The intelligence to implement and run these policies reside on the application delivery controller providing consistency and a non-labour intensive way of undertaking the desired level of management as and when needed.

Organizations need to also consider the consequences of sudden changes in the numbers of users accessing or using a particular application or website, malicious attacks on a service or slow web applications. These all have a dramatic effect on the levels of service an organization can deliver to its customers and increasingly its staff members. The application delivery controller from Zeus lets an organization define the service policies that deal with these problems and applies them easily and non-intrusively to application traffic, web services, types of transactions or groups of users for example.

Zeus Traffic Manager's Bandwidth Management capabilities are used to limit the amount of bandwidth that different services, types of transaction or groups of users can consume. For example, a low-priority service perhaps for staff members that hosts large file downloads could be limited so that it does not consume excessive bandwidth negatively impacting the level of service  you offer to customers, business partners or to other applications within the business that are deemed as more important to staff members.

Applications on the agenda

As the title of this post suggests, bandwidth management can be seen in a number of ways and will depend on where the individual sits within an organization and the types of applications and networks they deal with. As organizations deploy more applications and services internally to staff members and externally to customers and business partners, the emphasis to deal with these in a more intelligent manner in terms of the network bandwidth becomes even more crucial. Now is the time for network and application teams to come together to truly understand how each of their efforts can affect the others : )

For those of you who are interested in find out more about the Zeus approach to managing your bandwidth further information can be found here.

Mark Gyles

Optimizing your application for mobile device access

Short post this time inspired by a recent piece from Kristina O'Connell at Crescendo Networks titled 'Optimize Applications for Mobile Browsers'.

The piece highlights a Gartner report anticipating the growth in 2010 of handsets that will be capable of rich mobile commerce (Gartner's Top Ten Strategic Technologies for 2010). These handsets will look to access websites or other web applications and therefore as Kristina points out there will be a need to optimize your applications for this.

Kristina touches on the performance issues of applications being accessed by mobile handsets and offers good practical advice on how to achieve these optimizations citing data compression as one example, and how this can free up resources within your existing infrastructure.

However, at this stage it is important to add a further element to the discussion, but without delving too deep into the technical aspects: )  It is all very well and good optimizing your applications for access by mobile handsets/devices. But as we all know the number of handset/devices with different capabilities and frequency with which they come onto the market is quite amazing.

Therefore it is important that you optimize your applications accordingly for each different handset/device looking to access your website and web applications.

How can you do this?

Well, thanks to Luca Passani at WURFL (Wireless Universal Resource File) and his contributors, including individuals from a number of the carriers, there is a way! WURFL is basically a global database of all devices and their capabilities. As noted on the site, 'WURFL is an XML configuration file which contains information about capabilities and features of many mobile devices. The main scope of the file is to collect as much information as we can about all the existing mobile devices that access WAP pages so that developers will be able to build better applications and better services for the users.'

Luca and his many contributors also offer a Java API which is great news for those users of Zeus Traffic Manager who looking to undertake this type of application optimization on their application delivery controller.

Interested parties can find out more in an article by Zeus' Yvan Seth on our Zeus KnowledgeHub resource.

Mark Gyles

Bandwidth management but not as you know it...

This is part 1 of a two part post inspired by an article I found myself reading earlier this month titled ‘better understand bandwidth and your enterprise’s needs’. The article by Bruce Gain provides a practical hands-on approach to managing bandwidth for network admins who handle the network and the traffic through it.

Network layer 4

The article is a great example of how many organizations look at managing their bandwidth and how they might manage staff access and use of applications and web services above and beyond asking their ISP for larger pipes. The article and approaches contained within tends to resonate well with those viewing management of bandwidth from a network layer 4 traffic, WAN, network-admin, datacenter perspective : )

What about the application layer?

Well it's all very well and good talking about management of bandwidth at layer 4. But, actually when we start talking about how to ensure staff access and use of applications doesn’t have a negative impact on network bandwidth, we should be taking the discussion to the next level, the application layer of 7. At this layer an organization is able to use many more intelligent ways of managing the traffic to and from applications and manage how their use will impact or not impact the existing network infrastructure and resource.

How can this be achieved I hear you ask? Well in this situation an application delivery controller (application traffic manager, layer 7 load balancer, what-ever you wish to call it) can be used.

Other factors at play

Many organizations have applications running internally that take up network resource. Bandwidth is a key shared resource, and its misuse by one staff member could adversely affect the level of service that other staff members might experience.

At this point we should also consider two further elements that impact this discussion. Firstly, it may-be that unlike in the article which references ownership of dealing with the network bandwidth usuage residing with network admins, the onus may actually move to the applications teams whom have responsibility for the roll-out and use of applications on the network. We also need to think about the impact of staff access and usage of applications and services on the network, and how this might affect an organizations capability of offering services from an external facing web services point of view. The external facing web services may give customers and partners access to applications that are critical to doing business with them. Alternatively it may offer users/visitors the ability to buy goods and services over the web. An organization will need to seriously think about how they are going to ensure that the most valuable users get the best possible level of service from the resources available.

In part 2 of this post, available towards the end of this week, I will discuss how Zeus Traffic Manager (application delivery controller) can be used to implement and manage service policies. Service policies can deal with the issues that can cause bandwidth headaches for many IT professionals and organizations.

Mark Gyles

Zeus gets to wear a crown

Tuesday 17 Nov saw us gain further industry recognition for our solution, Zeus Traffic Manager. We were named by the Techworld Awards 2009 as Infrastructure Product of the Year. The solution was recognised for improving network and web-enabled applications by making them faster, more reliable, secure and easier to manage. The award assesses solutions that sit on the network including servers, routers, switches or accelerator/bandwidth shapers. 

So what, I hear you cry!

Well, everyone likes to gain recognition for what they do whether as an individual or a business and we're no different. This award adds weight to the fact that the businesses we talk to on a daily basis continue to see the importance of what our solution can offer them when they are looking to scale their networked application capabilities quickly. As a I am sure some of you will have experienced we place great emphasis on continually improving Zeus Traffic Manager through our R&D team. Whether you are looking for traffic management or load balancing capabilities in a physical, virtual or cloud environment makes no difference to us - you still get all the bells and whistles and great performance...

Other industry recognition and analyst reports can be viewed here.

Mark Gyles

How do I start using Zeus on Amazon Web Services

Following on from my earlier post where I mentioned that users of Amazon Web Services (EC2) could now use Zeus Traffic Manager on a pay-as-you-go basis. Well the folks in the R&D team have been busy always. Euan Harris in particular, has put together a series of FAQ's that will give you a good insight into how you can get started using Zeus on Amazon Web Services. I am sure these will be added to over time : )

The FAQ's which can be found on our technical resource site: ZeusKnowledgeHub cover the following topics:

• How do I configure my Zeus instance on EC2?

• How do I sign up to use Zeus AMI on EC2?

• How do I launch a new Zeus instance using Amazon's command line tools?

These FAQs are designed to give readers a step by step guide and we would encourage you to leave comments having run through them.

Mark Gyles

Pay-as-you-go traffic management on Amazon Web Services

Today saw the availability of Zeus Traffic Manager and Zeus Load Balancer on EC2 as Amazon Machine Images (AMIs). For those of you who don't know, an AMI is a template from which EC2 virtual machines, called instances, can be created. An instance is essentially a Zeus virtual appliance packaged for use on EC2.

Making the Zeus solution available on the EC2 platform on a pay-as-you-go hourly basis will make developing web services easier for developers and companies, particularly during the planning and production stages. By using a simple web service interface, users can monitor and control web traffic and automate maintenance to ensure online services are always fast and available even during high peaks in web traffic.

One recent example of this in action is with a UK company called HomeMade Digital. These guys are a digital service provider for a number of major UK charities. When HomeMade's clients need to respond to web traffic emergencies quickly, HomeMade need to be able to scale their IT infrastructure offering to provide the capacity to handle extra web traffic without any interuptions to service. HomeMade recently rolled out the full functionality of the Zeus Traffic Manager onto the Amazon Web Services Cloud. Adopting this approach has allowed them to keep data transfer and client hardware costs to a minimum.

If you would like to sign up to use Zeus Traffic Manager on Amazon EC2 you can find out more here.

Mark Gyles

Webinar: Read/Write Load Balancing for MySQL with Zeus Traffic Manager in the Joyent Public Cloud

Tuesday, November 17, 10am PST/6pm GMT

Raja Srinivasan from Zeus will be presenting alongside Rod Boothby, Joyent and Jimmy Guerrero, Sun Microsystems. This webinar will explore how to load balance MySQL in the Joyent Public Cloud with Zeus Traffic Manager and will demonstrate how Zeus Traffic Manager inspects, manages and load balances MySQL connections.

You can also expect a technical deep-dive into the best practices concerning high availability, performance and scaling in the cloud. If you are interested in learning more about how you can accelerate the development and deployment of MySQL-backed web applications, in the cloud, this webinar is for you.

To find out more and to register please click here.

Mark Gyles