Last weekend saw Lightning taking out datacenters associated with Amazon Cloud and Microsoft BPOS. It affected one Availability Zone (AZ) in AWS Europe. Rich Miller of Data Center Knowledge has detailed information on the incident.
Amazon said that lightning struck a transformer near its data center, causing an explosion and fire that knocked out utility service and left it unable to start its generators, resulting in a total power outage. While many sites were restored, Amazon said some sites that rely on one of its storage services may take 24-48 hours to fully recover. The company is bringing additional hardware online to try and speed the recovery process, but is advising customers whose sites are still offline to re-launch them in a different zone of its infrastructure.
Amazon said the event affected one of the EC2 Availability Zones in its Dublin data center, which is the company’s primary European hub for its cloud computing platform.
As it always happens with all the cloud outages, blogosphere and twitter were full of FUD about how one cannot trust cloud for their business. While there were many stories about startups getting affected, they conveniently ignored stories like Netflix using the same affected AZ for their Cassandra Cluster testing and still staying unaffected by the outage. As I pointed out in the article after April AWS downtime, any business using commodity public clouds should design for failure.
Do we have to use “Design for Failure” concept to use public clouds?
The answer is yes and no. Yes, because design for failure is essential to avoid disruptions due to cloud outages like the ones with AWS. If someone wants to take advantage of the cost savings (in certain circumstances) and agility offered by the commodity public clouds, designing for failure is imperative. It is the modern way of developing cloud scale applications. As I pointed out in my article, there are certain caveats which one should take into account while embracing this approach. As we go further and further into a cloud based world, we will see more and more applications built using this approach.
No, because you don’t necessarily have to use commodity public clouds. You may not realize the cost savings associated with commodity clouds but enterprise clouds offers better reliability while also offering the scale needed for enterprise applications. While many of the commodity cloud advocates may not agree with the idea of enterprise clouds, it is a necessary transitionary step in the cloud evolution. Simon Wardley has written a great piece on this topic and I recommend anyone interested in Cloud evolution to read it.
Whether we like it or not, Enterprise Clouds are going to stay relevant in the foreseeable future for following reasons:
- Not everyone like the idea of designing for failure. Especially, enterprise developers are still not ready to come out of legacy thinking. Even though developers are the flexible bunch among all of the IT practitioners, it takes time for their thinking to evolve. Right now we are seeing developers of social applications, developers in many startups, developers from cloud generation, progressive enterprise developers, etc. being bought into the design for failure philosophy. It will take quite some time before enterprise developers come around to this thinking
- Enterprises have invested lots of money and resources on legacy applications and they are not going to throw them away to develop greenfield apps. As long as such legacy apps stay in the picture, enterprise clouds are going to stay relevant
- Even though commodity infrastructure based cloud is growing in a big way, big chunk of money is still not there. Because of this, many service providers are still vary of embracing this model. Cloud platforms like OpenStack are showing tremendous promise but many service providers are still taking a “wait and see” approach. I have spoken to many datacenter folks and their first question (even now) is “will these platforms work with my existing VMware infrastructure?”. These folks will find enterprise cloud strategy more attractive, in the short run, than the commodity infrastructure based public clouds for their business.
As a public cloud advocate I hate this situation but, as an analyst, I am realistic about what is happening. It is going to take a long time before the tide changes and, till then, we are going to see the enterprise cloud buzz.
While on the topic of Enterprise Clouds, I want to highlight a news that came in today. Tier3, one of the enterprise cloud providers, today announced the Fall 2011 opening of two new Tier 3 data centers residing within Equinix International Business Exchange (IBX®) facilities in Chicago and Ashburn, Va. Tier 3 selected Equinix for the company’s enterprise-grade connectivity, global presence and carrier-neutral environment, which will enable Tier 3 to better meet the ever-increasing demand for secure, hybrid clouds in the enterprise. The data center in Chicago opens September 1, with the one in Ashburn opening in early Q4. In the aftermath of well publicized cloud outages, it is only natural for enterprise cloud providers to accelerate their datacenter and network capabilities.
Conclusion
Religious wars never end. Similarly debates around “technological religions” will not go away anytime soon. Whenever a change happens, there will be significant pushback from vendors whose business models are getting disrupted. Like the societies we live, technological change takes a long time and there will be many “transitionary approaches” filling the gap during this evolution. Enterprise clouds are one such solution to a problem many organizations are facing in their journey towards clouds. While it feels good to talk about physical reliability in the cloud, giving the option to build reliability into the applications is the right strategy for the future. If am an enterprise manager trying to map out my path to cloud nirvana, I would use enterprise clouds for the existing legacy applications while embracing the “design for failure” approach for newer apps. If you are an enterprise manager with a cloud strategy, I would love to talk to you. Please contact me.
This is the best summary of all those annoying twitter discussions (that I helped making even more annoying).
Massimo
With the benefits it offers, cloud service is aptly hailed as the next big thing for enterprise. But it comes with its own set of challenges, though.
For end-users, connecting to cloud servers requires costly high-speed Internet. Also, it may be essential to choose the kind of data we store on cloud. Some people may not be very comfortable having their medical records on it, for instance. And these outages may not happen every day, but they DO happen.
Cloud security is a huge issue as well. Remember when Dropbox left ALL accounts open for hours? It may not happen with Amazon but still somebody may try to hack it. A disgruntled employee can read sensitive information stored on cloud.
I truly wanted to construct a quick note so as to say thanks to you for some of the stunning steps you are writing here. My considerable internet lookup has now been rewarded with professional content to go over with my great friends. I ‘d point out that many of us visitors actually are unquestionably lucky to live in a wonderful website with many perfect people with insightful tips and hints. I feel quite privileged to have seen the site and look forward to some more awesome times reading here. Thank you once more for all the details.