Multi-Cloud is "The New Normal"

Not least the disaster of Nirvanix had shown that one should not rely on a single cloud provider. But regardless of spreading the risk, the usage of a multi-cloud strategy is a recommended approach which already is conscious or unconscious practionzed.

Hybrid or multi-cloud?

What actually means multi-cloud? Or what is the difference to a hybrid cloud? Well, by definition a hybrid cloud connects a private cloud or an on-premise IT infrastructure with a public cloud to supply the local infrastructure with further resources on-demand. This resources could be compute power, storage but even services or software. Is a local email system integrated with a SaaS CRM system it can spoken about a hybrid cloud. That means a hybrid cloud stands not just for IaaS or PaaS scenarios.

The multi-cloud approach extends the hybrid cloud idea by the amount of connected clouds. Exactly spoken, this can be n-clouds which are integrated in some kind of forms. Here, for example, cloud infrastructures are connected that applications can use different infrastructures or services in parallel or depending on the workload or the current price. Even the parallel or distributed storage of data over multiple clouds is imaginable to ensure the availability and redundance of the data. At the moment multi-cloud is intensively discussed in the IaaS area. Therefore taking a look on Ben Kepes‘ and Paul Miller’s Mapping Session on multi-cloud as well as Paul’s Sector RoadMap: Multicloud management in 2013 is recommendable.

What is often neglected, multi-cloud has a special importance in the SaaS area. The amount of new SaaS applications is growing from day to day and with that the demand to integrate this varying solutions and let exchange the data. Today, the cloud market moves uncontrolled into the direction of isolated applications while each solution offers an added value, but in the result leads to many small data silos. With this kind of progress enterprises already fought vainly in pre cloud times.

Spread the risk

Even if the cloud marketing always promise the availability and security of data, systems and applications, the responsibilty to ensure this lies in the own hands (This refers to an IaaS public cloud.). Although cloud vendor mostly provide ways and means using IaaS, the customer has to do things on its own. Outages known from the Amazon Web Services or unpredictable closures like Nirvanix should lead to more sensitivity while using cloud services. The risk needs to be spread consciously. Here not all eggs shut be put into one basked but strategically distributed over severall.


The best-of-breed strategy is the most widespread approach within IT enterprise architectures. Here multiple integrated industry solution for various sections within a company are used. The idea behind best-of-breed is to use the best solution in each case, an all-in-one solution usually cannot offer. This means one assembled the best services and applications for its purpose. Following this approach in the cloud, one is already in a multi-cloud scenario what means that approxiamtely 90 percent of all companies using cloud services are multi-cloud users. If the used cloud services are integrated remains doubtful.

Which is recommended on the one hand, is on the other side unavoidable in many cases. Although there are already a few good all-in-one solutions on the market. Nevertheless, most pearls are implemented independently, and must be combined with other solutions, for example email and office /collaboration with CRM and ERP. In respect of the risk this has its advantage, especially in the cloud. If a provider fails only a partial service is not available and not the overall productivity environment.

Avoid data silos: APIs and integration

Such best-of-breed approaches cloud marketplaces attempt to create, by grouping individual cloud solutions into different categories to offer companies a broad portfolio of different solutions, with which a cloud productivity suite can be put together.

Nevertheless, very highly controlled and supposedly integrated marketplaces show a crucial point with massive integration problems. There are many individual SaaS applications that do not interact. This means that there is no common database and for example the email service can not access the data in the CRM service and vice versa. This creates, as described above, individual data silos and isolated applications within the marketplace.

This example also illustrates the biggest problem with a multi-cloud approach. The integration of the many different and usually independent operating cloud services and their interfaces to each other.

Multi-cloud is “The New Normal” in the cloud

The topic of multi-cloud is currently highly debated especially in IaaS environment to spread the risk and to take advantage of the cost and benefits of different cloud infrastructures. But even in the SaaS environment the subject must necessarily become a greater importance to avoid data silos and isolated applications in the future, to simplify the integration and to support companies in their adaptation of their best-of-breed strategy.

Notwithstanding these expectations, the multi-cloud use is already reality by companies using multiple cloud solutions from many different vendors, even if the services are not yet (fully) integrated.


Nirvanix. A living hell. Why multi-cloud matters.

One or two will certainly have heard of it. Nirvanix has oneself transported to the afterlife. The enterprise cloud storage service, which had a wide cooperation with IBM, on September 16, 2013 suddenly announced its closure and initially granted its existing customers a period of two weeks to migrate their data. The period has been extended to October 15, 2013 as customers need more time for migration. As a Nirvanix customer reported, it has stored 20 petabytes of data.

The end of Nirvanix

Out of nowhere enterprise cloud storage provider Nirvanix announced its end on September 16, 2013. To date its not declared how it happened. Rumor has it that a further round financing failed. Other reasons are seemingly on the faulty management. Thus, the company had five CEOs since 2008 until today. One should also not forget the strong competition in the cloud storage environment. Firstly, in recent years, many vendors have tried their luck. On the other hand the two top dogs Amazon Web Services with Amazon S3, and Microsoft with Azure Storage reduce the prices of their services in regular cycles, which are also enterprise-ready. Even to be named as one of the top cloud storage service provider by Gartner couldn’t help Nirvanix.

Particularly controversial is the fact that in 2011, Nirvanix has completed a five-year contract with IBM to expand IBM’s SmartCloud Enterprise storage services with cloud-based storage. As IBM has announced, stored data on Nirvanix will be migrated to the IBM SoftLayer object storage. As an IBM customer, I would still ask carefully about my stored data.

Multi-Cloud: Spread your eggs over multiple nests

First, a salute to the venture capital community. If it’s true that Nirvanix had to stop the service due to a failed round financing, then we see what responsibility is in their hands. Say no more.

How to deal with such a horror scenario like Nirvanix as cloud user? Well, as you can see a good customer base and partnerships with global players seems to be no guarantee that a service survived long term. Even Google currently plays its cloud strategy on the back of its customers and makes no binding commitment over the long-term consist of its services on the Google Cloud Platform, such as the Google Compute Engine (GCE). On the contrary, it is assumed that the GCE will not survive as long as other well-known Google services.

Backup and Multi-Cloud

Even if the cloud storage provider has to ensure the availability of the data, as a customer you have a duty of care and must be informed about the state of your data and – even in the cloud – take care of redundancy and backup. Meanwhile functions in the most popular cloud storage services are integrated to make seamless backups of the data and create multiple copies.

Although we are in the era of cloud, yet still applies: Backup! You should therefore ensure that a constantly checked(!) and a reliable backup and recovery plan exist. Furthermore, sufficient bandwidth must be available to move the data as soon as possible. This should also be checked at regular intervals using a migration audit to act quickly in the case of all cases.

To just move 20 petabytes of data is no easy task. Therefore you have to think about other approaches. Multi-cloud is a concept which is gaining more and more importance in the future. At it data and applications are distributed (in parallel) across multiple cloud platforms and providers. On this my analyst colleagues and friends Paul Miller and Ben Kepes already had discussed during their mapping session at the GigaOM Structure in San Francisco. Paul subsequently had written an interesting sector roadmap report on multi-cloud management.

Even next to Scalr, CliQr, RightScale and Enstratius already exist some management platforms for multi-cloud, we still find ourselves in a very early stage in terms of use. schnee von morgen webTV by Nikolai Longolius for example, is primarily on the Amazon Web Services and has developed a native web application 1:1 for the Google App Engine as a fallback scenario. This is not a multi-cloud approach, but shows its importance to achieve less effort for a provider-cross high availability and scalability. As Paul’s Sector Roadmap shows, it is in particular the compatibility of the APIs that must be attributed a great importance. In the future companies can no longer rely on a single provider, but distribute their data and applications across multiple providers to drive a best-of-breed strategy and to specifically spread the risk.

This should also be taken into consideration when simply store “only” data in the cloud. The golden nest is the sum of a plurality of distributed.