Always a content-rich conference, this year’s show does not disappoint. I look forward to all the cloud announcements that have already begun to come out – and just like customers I face the same challenge of separating the “wheat’ from the hype/chaff. A new game – guess how many companies will announce themselves (new and reinvented) this year with some form of “cloud” or even better “clou” in their names…

Back to VMworld – as I listen to Ed  Bugnion from Cisco plug UCS in the background – this year, VMware’s theme is Virtual Road, Actual Clouds. As opposed to the 3 or 4 actual case studies at the show last year, more customers with public, private and hybrid cloud deployments are here this year and ready to tell their ROI stories. An example from the first session I’m in is BT, which went on an 8-month virtualization (maybe cloud) journey and here were the results:

• from 700 racks at 8 sites to 40 racks at 5 sites
• from $2.4 million spent anually for power to $110K
• from 9309 network ports to 840
• from weeks to provision to 12 minutes
• from 96 hours to backup to about half an hour
• from onsite management required – to fully remote management enabled

Pretty impressive – but to be taken with a grain of salt since BT is of course a vendor delivering virtualization/cloud services themselves.

But more than ROI and case studies, another theme is emerging which makes me feel like I’m back at a Gartner show. Service-oriented architecture. IT as a Service – ITaaS. Which comes first – the analyst or the big vendor messaging?

More on this tomorrow from the VMworld keynote session. Register to see it streamed live.

(Photo Credit: The Creativity Works, VMworld 2008)

Here are a few recent news stories around virtualization and the cloud which makes us wonder whether or not these issues will be addressed at VMworld.

How virtualization is affecting server sales. Recent articles point to more and more customers buying integrated infrastructure packages rather than individual servers. Since virtualization is decreasing the number of physical servers needed to run a data center, these packages are including not only servers, but storage and networking components.

What’s next for virtualization? Obviously virtualization management and cloud management go hand in hand, but what else can we expect to see in the virtualization industry? Daniel Beveridge recently wrote a post which pointed to the up and coming trends for virtualization. He mentions the birth and growth of Virtual Desktop Infrastructure (VDI), the increase in different hypervisor types, the case for extending storage and the  “depersonalization of virtualization”. He also talks about “mist computing” – a model where virtualization spans the space between data center and desktop. Compute resources that are all around us, nearby and omnipresent. According to Beveridge, mist computing is the future of virtualization.

What VMware announcements are coming? Some speculation has us wondering what big news is coming out of VMworld this year. Most large industry tradeshows (like MacWorld, Comic-Con or CES) are focused around major product announcements or industry changing trends. Since Sphere 4.1 was recently released, it leaves one to wonder what other big announcement is coming.

The future of VMware. Information Week posted a couple interviews with VMware CEO Paul Maritz. VMware has decided to escalate its virtualization strategy far beyond the world of IT infrastructure and into the areas of application and user-device virtualization, while also dramatically fortifying its capabilities in end-to-end IT management. VMware is pushing a three-level strategy for stringing together infrastructure, applications, and end-user devices in a way that helps customers save not only on CapEx but also on OpEx—and beyond. It looks like VMware is searching for “unified and harmonious management" over data centers, IT infrastructure and applications.

• The product (EM7) that ScienceLogic designed 7 years ago is a great fit to help IT operations manage their infrastructure in public cloud, build a private cloud and hybrid cloud, as well as manage their traditional IT operations infrastructure.
• Specifically, when we think about the cloud, our customers tell us that there is a series of applications and solutions they have that make sense to be in a variety of locations. Part of it is scalability, part is economics, part is the nature of the application.
• The challenge they are facing is having one consistent, consolidated view.
• When we started the company 7 years ago, the industry was very siloed – network  management in one, security management in another, application management in another.
• We wanted a solution from the beginning that offered a multi-tenant view. We wanted to deliver as a service. We wanted to be a silobuster and provide a complete view to deliver as a service.
• To integrate infrastructure as service, you still want a complete view. One view. No matter where the infrastructure is hosted. You want one view of how that infrastructure is behaving.
• Cloud computing is not just a big deal, it’s a generational thing. We see that in our business. The next generation of technology leaders are looking for innovative, fast ways to deliver meaningful services – for B2B, B2C or internal service offerings within an enterprise.
• Need for agility and speed to give customer what they want – it’s driving IT to change the way they use tools to deliver those services.

If you’re still confused about cloud computing, check out this post on Network World that provides the key concepts of cloud computing. The Vancouver Sun also shares some basic tenants of Platform as a Service (SaaS), Infrastructure as a Service (IaaS) and perhaps the most familiar, Software as a Service (SaaS).

Advantages and disadvantages of private cloud are examined in CIO News from TechTarget. Three main pitfalls to avoid are outlined, including expecting speed and agility, waiting to get in the game, and failing to understand self-service (from the end user and cloud administrator) and true production needs. Need more persuasion toward private cloud? According to the article, “once private clouds are up and running, they are paying for themselves in one to two quarters. An estimated one in 10 enterprise servers are sitting idle; private clouds enable them to move from less than 10% server utilization to 70%.”

A new personal blog from VMware developer Steve Jin points us to system provisioning in cloud computing, a two part post. OS, middleware, and application tools for provisioning are discussed. Steve says that in a cloud application architecture, “you want to design your applications to be as stateless as possible so that you can standardize the application for massive deployment.” He provides some examples of application provisioning tools, and summarizes saying, “System provisioning in cloud computing is an important aspect of system architecture, application design and operation, the standardized templates provide not only operational efficiency but also high quality, pre-qualified software stacks for building cloud applications.”

Although cloud computing could save the government millions of dollars, 22 out of 24 major agency executives are still worried about security, according to a GAO report. But, it appears there are more benefits than risks, at least according to Mike Bradshaw, director of Google Federal:

Despite the concerns, cloud computing will improve security. Federal. Cloud computing vendors store data on multiple servers in multiple locations, making it difficult for cybercriminals to target one location. The redundancy also means agencies are protected against disasters. The cloud enhances security by enabling data to be stored centrally with continuous and automated network analysis and protection.

While some IT system and network administrators may worry that the cloud will put them out of the job, GovinfoSecurity says otherwise. When it comes to IT security professionals need to be prepared for the cloud and for changes in how IT operations management is managed. Practitioners should focus on how the technology enables the delivery of new and unique business capability. Think about a hierarchical list of applications, environments and/or infrastructure that could be replaced by commodity cloud services, and work with management to understand how IT services can be tied together to business processes.

It is really a couple of years since Cisco initially laid out these bets on video, collaboration, cloud, and smart and connected communities to lead to the next big Internet transformation. It is clear that we are well down the path to transformation with the first three of these, and Cisco’s bets are beginning to pay off handsomely with video (all the way from the Flip camera to large corporate Telepresence rooms) and UCS sales beginning to take off as a cloud enablement solution. Padmasree emphasized Cisco’s impact on cloud services at all levels, including their own SaaS applications: “Cisco Hosted Collaboration Solution” EOS, Scansafe, Postpath, Telepresence as a service, Webex, and IronPort. It was truly gratifying to know that EM7 is already delivering service assurance for Cisco in two of the seven SaaS applications listed by Cisco’s CTO (Telepresence and IronPort). Now for the other five!

The Smart and Connected Communities is a little longer term bet for Cisco, but with a potentially huge payout. With mass urban migration in populous developing countries such as India and China, there is a projected need for hundreds of new cities over the coming decades. Cisco’s vision is to be at the heart of this new development with new cities built “pre-networked”. Consumer Telepresence, smart grid, physical safety and security all play key roles in the product strategy for this, and already New Songdo City in South Korea is being built as a prototype for what is possible. With our close collaboration with Cisco on Physical Safety and Security, we hope to see EM7 play a major role in this transformation.

During the Advanced Cisco IOS Device Instrumentation session here at Cisco Live! It became clear that all of these and much more are now possible directly from within IOS.

Much of this power comes from the recently added ability to run TCL scripts directly on the device. It turns out TCL has been in IOS for a long time, it was used in delivery of many built-in IOS features, but was not available for access by end users. Now however, this feature is unlocked and network engineers can go to town writing TCL scripts triggered by events, or potentially scheduled from Kron (yes, that’s cron with a “K”). The TCL language is a convenient tool for pattern matching against events and performing local IOS operations, or generating enterprise specific events “Northbound” to an NMS such as EM7, but it comes with some risk. Pattern matching can be CPU intensive, so administrators will need to take care not to overload the device with custom code and limit its ability to perform its primary function of routing and switching. Actually IOS includes some mechanisms to limit the impact of custom code by implementing thresholds for how long a script can take to run, and also for how much CPU it can consume.

The TCL scripting language is not the only way that administrators can implement sophisticated logic on the IOS platform. The Embedded Event Manager (EEM) has been getting lots of love and attention from the Cisco Engineers as well. The EEM applets now support logical constructs such as loops and conditional statements. As of EEM 3.1, the tool can intercept SNMP GET and SET operations and take custom actions for particular oids. This is a great way to extend the SNMP capabilities of IOS with your own custom MIB, just has been possible in the server world for many years with extensible agents like net-SNMP.

These new IOS capabilities are powerful new strings for the network engineer’s bow, but they will need to consider carefully when to use them. Cisco network monitoring is made possible with systems such as EM7 that provide the ability to perform pattern patching and sophisticated event handling outside the network device, with full audit trail and logging built in, and with consistent operation across the network. Custom event handlers running on IOS itself have the same considerations around logging, audit trail and version control that apply to other software applications, using these in an uncontrolled way could be a recipe for disaster. Undoubtedly these tools have their place though, and will be invaluable to taking local actions immediately in response to specific IOS events.

Today Cisco really began the process of demonstrating service orchestration, using partner solutions to deliver orchestrated service activation, provisioning, capacity management, and finally service clean-up. The first partner that Cisco has walked this path with is BMC. Through the BladeLogic acquisition and Atrium Orchestrator, on paper BMC has a great core tool set for cloud service orchestration.

Unsurprisingly the Q&A session revealed things are not quite that simple. There are an awful lot of moving parts in the BMC pieces alone, with many of the components coming from different acquired companies, and each product having its own nuances and unique architectural slant. To get network and UCS orchestration to the point where reference templates could be delivered for some specific Cisco devices required a small army of BMC consultants. What’s more, there is no solution for storage provisioning in this configuration yet. What may be even more surprising is that BMC has nothing around orchestration of cloud service assurance, something that we offer in EM7 today through the RESTful APIs on the EM7 Integration Server.

This was a first foray into end-to-end cloud orchestration for Cisco, and undoubtedly solutions in this space will evolve and mature rapidly. For now though, service providers investing in this area need to do so with their eyes wide open and full awareness of the (lack of) maturity of solutions in this area.

As the official monitoring and helpdesk provider (third year running!), we provided crucial cloud monitoring, network monitoring and help desk functions for InteropNet, which this year, was run like a private cloud.

We used EM7 G3, last year’s Best of Interop winner for network management, to enable and manage the InteropNet hybrid cloud environment (multi-data center + private cloud + some public cloud deployment). Beyond the traditional network management and help desk functionality we’ve been providing InteropNet for 3 years now, EM7  G3 provided virtualization monitoring (heavily virtualized InteropNet this year with many vendor solutions delivered virtually). For both physical and virtual devices, EM7 G3 acted as a single consistent monitoring solution and MoM, providing a “single pane of glass” view across all components of the InteropNet infrastructure. To support the private cloud aspects of InteropNet, EM7 provided multitenancy -  monitoring data segregated for each “customer” (for InteropNet, the “customer” was each exhibitor) in this shared resources environment and chargeback based on actual network bandwidth usage data at a “customer” level – for a true “pay by the drink” billing capability (although no one was actually presented with a bill).

One more cool thing – we worked with Citrix which performed wide-area load balancing of the EM7application servers available on-premise as physical servers and in the Opsource cloud as a virtual appliance.

As the world’s largest temporary network, it’s great that InteropNet was ready for the cloud. In a real world environment, these features have many benefits towards running a true private cloud. Chargeback is especially important for enterprise companies and government agencies,  because it allows them to apply standard rates and charge based on usage. Service providers have been doing this for years, and with an increase of users searching for the perfect private cloud, they are well ahead of the game.

There is a tremendous amount of awareness according to a Oracle survey. 44% of large enterprises are interested in building an internal cloud. This shouldn’t be surprising. A public cloud is still largely driven b y greenfield customers  – but an internal cloud and private cloud is driven by people with large problems today.

IT supporting the business vs.  IT as a cost center. Cloud infrastructure maximizes return and uses technology as a competitive weapon. It provides you with scale,  optimized systems, performance, openness, speed to implementation and massive data capabilities.

We talk again about the enterprise evolution to cloud – moving from silo’d solutions to the grid. Moving to grid is one of the first critical steps, moving to a private cloud that allows for self-service, chargeback, capacity planning and policy-based resource management.

There are three steps to IaaS. Use virtualization as IaaS Enabler. Putting things into virtual machine containers is the first step. They need to be isolated to look at how they are packaged and what the requirements are. You need the basic ability to deploy multiple copies of the application. Cloud is not just server virtualization. With grid computing and clustering, we want many resources to look like one.

Traditional virtualization vs. virtualization for clouds

• Traditional virtualization – VM provisioning. Virtualization for cloud – library of virtual appliances/assemblies for rapid, low-risk VM creation and deployment.
• Traditional virtualization – Resource management and HA – no dynamic resource management/automation. Virtualization for cloud – policy automation to dynamically load-balance VMs across / between pools of  resources
• Traditional virtualization – storage and network – consolidated servers with local / direct, static and inflexible connections. Virtualization for cloud – shared storage, flexible networking to create server pools.
• Traditional virtualization – admin / operations model – traditional (sysadmins, storage admins, network react to tickets to provision VMS, storage, network). Virtualization for cloud – automation supports self-service provisioning of VM with administrators proactively managing general pool  scaling.

Elastic infrastructure

Release engineering – virtualization highlights deployment processes. You have to consider  application packaging and deployment, orchestration and process management (what are the direct business results. IT as an enabler, not a cost), performance monitoring, and pre-requisite for public, private or hybrid cloud usage.

Balanced systems  are key.  A balanced system deliver both performance and mission-critical functionality (at the same time).

Elasticity – a fundamental cost driver. 24×7 use of cloud is likely more expensive.Elasticity is a strategy. Elasticity within the data center helps to drive efficiency. Demand and capacity are colliding – and clouds are in the middle. We’re talking about demand of users, services and access. And capacity issues requiring power, costs, space and heat.

Final food for thought – a recent online thread about cloud computing had people saying systems administrations is dead – you don’t need it in the cloud. Actually it’s harder. You need to look at key business requirements and SLAs – how do you keep them in a virtualized world. We have a much more agile data center, and have to manage physical and virtual worlds.

Operational efficiency we think about monitoring, autonomic scaling, automation, self-service portal, policy management, resource management and policy enforcement.

Deployment efficiency – we think about a traditional three tier architecture (database tier, app tier, web tier) – when we move to the cloud we think of a one-to-one relationship. What do you do when you consider a load balancer and HA? We are seeing customers build appliances. We think an idea that will be important is an idea of assembly in a much more cost effective and scalable manner.

With runtime efficiency, we see people focusing on back-end systems. One of the  key bottlenecks we’ve found is the inability to put a lot of Java into a single machine. The challenge becomes, how to put multiple JVMs in place across different machines? The problem is similar to trying to put rocks in a bucket.  You have a certain size, and you have to fit a little rocks and big rocks. The same with containers – you have only so much space in a JVM, which then means you need to have multiple JVMs. We are now running JVMs virtualized. This makes it extremely efficient not only for performance but for planning.

Which cloud model are people primarily considering? Hybrid cloud with 96% – according to a survey by Oracle.

Operate your IT organization as a cloud. Think about it – private cloud is real and delivers more than hardware savings. They are practical and deliver results – 35% reduction in operating costs. PaaS delivers the highest return for the enterprise. Standardized platforms deliver fastest time to market, lowest cost and best quality.