If everyone is thinking the same, someone isn't thinking

Lori MacVittie

Subscribe to Lori MacVittie: eMailAlertsEmail Alerts
Get Lori MacVittie: homepageHomepage mobileMobile rssRSS facebookFacebook twitterTwitter linkedinLinkedIn

Related Topics: Cloud Computing, Virtualization Magazine, Cloud Interoperability, VMware Journal, Infrastructure On Demand, Cloudonomics Journal, Intel XML, XML Magazine, SEO Journal, Infrastructure 2.0 Journal, SOA Best Practices Digest, Datacenter Automation, SOA & WOA Magazine, SOA in the Cloud Expo

Blog Feed Post

The Cloud Metastructure Hubub

How Infrastructure 2.0 might leverage publish-subscribe technology like PubSubHubub to enable portability of applications

Tower of Babel by Pieter Bruegel the Elder.
One of the topics surrounding cloud computing that continues to rear its ugly head is the problem of portability across clouds. Avoiding vendor lock-in has been problematic since the day the first line of proprietary code was written and cloud computing does nothing to address this. If anything, cloud makes this worse because one of its premises is that users (that’s you, IT staff) need not concern themselves with the underlying infrastructure. It’s a service, right, so you just use it and don’t worry about it.

Let’s assume for a moment that you can easily move applications from data center to cloud to cloud. Plenty of folks are working on that, but very few of them address the “rest of the story”: the metastructure.

Metastructure contains the metadata that describes the network, application network, and security infrastructure providing all those “don’t worry about” services cloud providers offer. Load balancing, firewalls, IPS, IDS, application acceleration, secure remote access. If you’ve spent time with your cloud provider tweaking those services – or configuring them yourself – then moving to a new cloud provider is not only a huge investment in time, it’s actually going to be painful because you’re essentially going to have to recreate every metastructure configuration again.

Yes, you’ve done this inside your own data center for years. Every forklift replacement or upgrade of infrastructure has come with its own load of baggage in the configuration arena. Switching out vendor equipment – especially core components – can be extremely painful, especially when configurations need to essentially be “translated” between them. But cloud makes this worse because technically speaking you don’t even have access to the existing configurations. You can’t see them, you can’t have them, and you can’t run them through whatever “upgrade” or “migration” script your new vendor offers to ease the process.

Are you depressed yet?

There’s been some talk of including metastructure data with the virtual machine, but the problem with this is that it almost always requires that the meta data be wrapped up using a proprietary API, such as is provided by VMware. That’s okay if you restrict yourself to only cloud providers that use the same virtualization technology, but not okay if you want to be able to make a move from one technology to another. It also assumes that the metadata is specific to the infrastructure, which is even more unlikely when moving between cloud providers.

HOW ABOUT A CLOUD-BASED CMDB (Configuration Management Database)?

There are several ongoing efforts to address this very scenario because it is so painful. Most of them would, if adopted, require vendors to implement support for a specific standard so that configurations can be managed and exchanged in that standard format. That makes sense, that’s how we’ve always handled translation of data between disparate systems that don’t speak the same language. In the application world we call the process of mapping one format to another “integration” and you can easily evoke a look of terror on a co-worker’s face just by saying the word within their range of hearing. Go ahead, try it. Just make sure they aren’t carrying anything heavy that can be easily thrown at you when you do.

CMDB (Configuration Management Database) technology is another method of addressing the problem of, well, managing configurations. These solutions store configuration of a wide variety of infrastructure solutions – from routers and switches to web and application servers to application delivery controllers. They do a great job of managing configuration and can even “push” configuration out to devices if so desired. But the configurations stored and managed in a CMDB are product-specific, not generic, so they can’t adequately today address the problem of portability.

You can probably see where this is going: a cross between CMDB and a nice, industry-wide standard would probably do the trick, wouldn’t it? And if it was public (in the sense that any application or service is public on the network – that is, accessible via the Internet to any cloud provider or customer site) then cloud providers and organizations alike could take advantage of that configuration management mechanism and use it to their advantage. Portability becomes possible rather than fantasy.


Cloud providers and organizations alike are likely to stop right there. Sharing configuration of infrastructure and core components is just asking for trouble. If ever such a cloud-based CMDB were compromised, well…let’s just say it would be A Very Bad Thing.

But what if the actual metadata, the configuration information, were stored either in the enterprise or the cloud provider (or both), and merely pushed and pulled via a public mechanism on-demand?  Configuration isn’t changed all that often and if an organization is moving between clouds they certainly know when they’re doing it. If there was some mechanism through which metastructure could be published and to which infrastructure could subscribe then when changes were made or providers changed that metastructure data could be easily grabbed from the public cloud-CMDB system (cloud catalog, anyone?) and interpreted into product-specific configuration by the products themselves.

Think of  it like SOA clients pulling WSDL (Web Services Description Language) from a UDDI (Universal Description, Discovery, and Integration) server. The SOA client pulls the WSDL, which describes the service(s), configures itself appropriately, and then is able to make use of those services. The intent of introducing UDDI was a service-catalog that could be polled on-demand to provide the latest information about the service and describe it in an abstract, vendor-neutral way such that any client could access any service, regardless of implementation language or environment. Sounds a lot like what we want for infrastructure portability, doesn’t it?

PUBSUBHUBHUBThat’s where PubSubHubub comes in. While this draft standard for a publish-subscribe system is generally being leveraged by software developers to enable faster sharing of information across the Internet, it is also a fine example of a system that could be used by infrastructure 2.0 solutions to share metastructure. Consider the existence of a public PubSubhubub Hub, like Google’s public PubSubHubub Hub, and how it might be leveraged to share metastructure between clouds or the organization and the cloud.

Note that XMPP is used today by at least one cloud provider to enable distributed cloud management in a nature very similar to that of PubSubhubhub.

In any case, the specific implementation of the configuration “hub” is relatively unimportant; what’s important is that (a) customers can publish a vendor-neutral metastructure to an isolated channel that communicates their specific infrastructure needs and (b) providers can subscribe, at will, to customer topics and retrieve metastructure in a way that allows their infrastructure to in turn configuration itself (or be configured by the provider’s system, as is required by the provider’s implementation).

Early on it would be necessary for the cloud provider to provide the “translation” and configuration services simply because even if a metastructure standard existed today (and it doesn’t) it would take months and possibly years before all the possible infrastructure vendors were able to update their systems to interpret the standard. If the provider implements a configuration “gateway”, however, he can immediately take advantage of such a standard and use existing skills and knowledge gained from its automation and orchestration of its cloud to configure the infrastructure appropriately based on the metastructure. This has the added advantage of “hiding” the infrastructure implementation from the outside world, which for some providers is a very important thing to do.


That’s okay for two reasons: first, we ensure that the metadata description is XML-based, because it’s extensible. If we build into the standard a way to extend it naturally such as is provided with XML the interpreters (configuration “gateways”) can either (a) translate if it can or (b) ignore.

Consider the use of OVF (Open Virtualization Format) to further describe what is called a Virtual Machine Contract (VMC):

For each virtual system, the associated metadata is described in a set of specific sections. The VirtualHardwareSection describes the virtual hardware required including the amount of memory, number of CPUs, information about network interfaces, etc. The OperatingSystemSection describes the guest operating system that will run in the virtual system. The ProductSection provides basic information such as the name and vendor of the appliance and can also specify a set of properties that can be used to customize the appliance.

While VMC is very basic at this point, it’s a good start at providing the foundation for building out a more complete, standards-based description of the metastructure necessary to configure an infrastructure to deploy a specific application in a virtual machine format. Using this as the basis for metadata exchange – when fully described – via a public hub could alleviate most of the issues with sharing infrastructure metadata (metastructure) across clouds in a generally vendor non-specific manner. In other words, portability of both the virtual machine and the specific infrastructure configurations necessary to optimally execute and deliver the application to the end user in the most fast and secure manner possible.

We’re nowhere near this point, by the way. VMC needs to be fleshed out as far as standard metadata goes for infrastructure (perhaps a good chore for the SRI Infrastructure 2.0 Working Group) and vendors would need to adopt and extend out the ProductSection of VMC for product specific configuration that isn’t included in the base format. And PubSubHubub would need to be proven to be a secure method of exchanging the metastructure across clouds. What is likely is that as we move forward trying to extend the plateau of collaboration down the stack toward the core infrastructure is that a new set of tools, products, solutions, and services will emerge to fill the unavoidable gaps in the standards, e.g. a service-based cloud configuration hub offering translation of proprietary metastructure data to some other proprietary metastructure data.

Perhaps there’s a better way overall, and OVF/VMC and PubSubHubub will simply remain in our memories as the catalyst and template for a different set of standards providing portability across clouds. But there is a way to provide this level of portability and collaboration across clouds, across the infrastructure and the application. The need – and perhaps more importantly the belief that it’s necessary to address the need – is growing.

UPDATE: Christofer Hoff pointed out that vCloud has been submitted to the DMTF for standardization, technically making it "open" rather than "proprietary." It is still only implemented by VMware technologies, so for the time being it might as well be proprietary, but this may change in the future.

Follow me on Twitter View Lori's profile on SlideShare friendfeedicon_facebook AddThis Feed Button Bookmark and Share

Related blogs & articles:

More Stories By Lori MacVittie

Lori MacVittie is responsible for education and evangelism of application services available across F5’s entire product suite. Her role includes authorship of technical materials and participation in a number of community-based forums and industry standards organizations, among other efforts. MacVittie has extensive programming experience as an application architect, as well as network and systems development and administration expertise. Prior to joining F5, MacVittie was an award-winning Senior Technology Editor at Network Computing Magazine, where she conducted product research and evaluation focused on integration with application and network architectures, and authored articles on a variety of topics aimed at IT professionals. Her most recent area of focus included SOA-related products and architectures. She holds a B.S. in Information and Computing Science from the University of Wisconsin at Green Bay, and an M.S. in Computer Science from Nova Southeastern University.

Comments (0)

Share your thoughts on this story.

Add your comment
You must be signed in to add a comment. Sign-in | Register

In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.