Posted on 11 July 2012
I live in the Washington, DC metro area, specifically the 51st state – Northern VA or better known as NoVA. Fortunately, I was not affected by the storms and power outages of a couple weeks ago. However, Amazon and their Ashburn data center once again. Once again, many of us social media users found the pain of the outage.
How many outages do we have to endure before these folks put together better monitoring and DR plans? The outage begs the question of how important are these sites in the grand scheme of things as it relates to human life, power outages, tainted water and housing destruction during a storm. My response is all in how you look at the situation. One way to look at it is this is their business and livelihood, so DR should be relevant. Another is that many of us use these sites during times of destruction to communicate with the rest of the world, not just for fun or business.
As a marketer and business persoin, I find it interesting none of them have found it a competitive advantage to have the best plan and be alive when their competition is down as Okta exclaimed over this last outage, but security is their business and they must be alive at all times to retain customers.
I’m excited about all of the change with technology underfoot, but control and flexibility must be weighed.
How are you preparing to manage your mixed environment?
Travis just posted, “Unlike the Weather, You Can Choose Your Cloud”, discussing the recent power outages we experienced here in the Northern VA area last week. As a native of this area, thunder storms and hurricanes do pass our way and affect us during this time of year. Late May of 2008, we had a storm very similar to this one and this isn’t the first Amazon outage, “Amazon EC2 Outage Downs Reddit, Quora“ April 2011, nor will it be their last. Intuit’s SaaS QuickBooks was down 36 hours in June 2010, “Update: Intuit Sites Outage Strands Thousands of SMBs“. Or just this past Tuesday Salesforce.com was down, triggered by a power outage at an Equinix data center in Silicon Valley. Each time one of these outages occurs, it gains big headlines that the cloud has failed. (Read more here …..)
Posted on 24 April 2011
The Hub Commentary_
The news continued through the weekend on the Amazon outage this past week. While many who are commenting are using it as an opportunity to point the finger that the cloud has failed. I do not. The cloud is here to stay.
The role of the IT organization is changing to that of the service provider managing the services delivered to the business. Just because a public cloud becomes the deployment option of choice does not let IT off the hook for responsibility of the service. Think about it as your car and taking it in for service. Just because Ford, GM, etc. didn’t ship a part doesn’t make you any less happy when the dealer says, “I didn’t get the part from the manufacturer”. Did you take your car to the dealer or the manufacturer?
Agile and dynamic infrastructures are here to stay. Those that embrace them and take advantage of them early, will lead their industries. As one article stated that one of those affected said, “we wouldn’t be as far along without Amazon”. What this boils down to is what often lags with new technology – management.
I wrote a previous article on defining service value and then define deployment options, service levels, back-up and recovery. The natural market reaction will be to impose stringent service level requirements on the service providers, I’m here to tell you that you will pay for doing that. Define the appropriate service level for the criticality of the service. Then I suggest planning for the contingency of a major outage, don’t pay for the risk of the catastrophic every day for something that may happen once, never, etc. Have a back-up plan.
What is your back-up plan?
Despite restoring service for many customers, Amazon’s troubles with its Elastic Compute Cloud (EC2) and Relational Database Service continue this morning, meaning that debate over the wisdom of relying on such cloud services is certain to grow louder. (Read Full Article…)
Posted on 21 April 2011
The Hub Commentary_
Outages always make for big headlines and make for good examples of the cost of visibility and management. I read these articles with positive thoughts and still see cloud computing as the future and those who are bold took advantage of growing business earlier rather than later.
As noted in the first article, some of these organizations would not be where they are today with their business if they had not leveraged new deployment options as fast as they did. In a previous post, I discussed mapping your services based upon business value and cost and how to manage those services and selecting deployment options. It’s a balancing act of how bold and how much of a risk you take to grow your business with cutting edge technology. The balance to strike is how much you spend to have a DR or back-up plan for an event like this as you know they will happen.
Outages will occur and this shouldn’t push folks to write unwieldy service level agreements as you pay for those by pushing the risk back to the provider. I would suggest that spend would be more wisely spent on a back-up provider to account for hiccups in a cost effective manner.
My hat still goes off to those embracing the cloud and growing their business as a result with agile technologies and there are learnings to plan for moving forward.
How are using agile technologies?
A partial failure at Amazon Web Services’ cloud-computing infrastructure brought down some Internet operations today, including the Web sites of Quora and Reddit. (Read Full Article…)
Amazon reports this morning that it is making progress in restoring full service to customers of its Elastic Compute Cloud (EC2) and Relational Database Service in the eastern portion of the country after a rocky stretch of trouble that began sometime before midnight. (Read Full Article…)
Posted on 18 April 2011
The Hub Commentary_
I would agree network performance is an aspect to consider when setting up and considering deployment options for mass storage. Another aspect to consider is something I covered in another post on Categorizing your Serivces and which services, inclusive of data, should be put out in the cloud.
In all sourcing situations you need to consider: security, performance, responsiveness to recovery, and management of the vendor. A good rule of thumb estimate in managing the provider is 3-7% of the value of the contract. This is an often overlooked cost in these short sided decisions to reduce costs short term by removing hardware, management and staff potentially.
Storage is well suited for cloud services, however, the service it supports and the criticality to the business is a high consideration. Commodity data and services are easily the first to be considered for external service providers. However the more mission critical and data sensitive it becomes to the organization, the more cost prohibitive it becomes to source the service.
Yes, performance is a key to consider, however, I still surmise that the service criticality to the business must also be in the decision process.
How do you decide how to right source your IT services?
We all know that the cloud can provide cheaper storage than in-house infrastructure, but is it better?
Of course, there are numerous ways to quantify “better,” but from an operations standpoint, it wouldn’t hurt to know exactly what you’re getting and what you’re giving up when you sign on the dotted line. (Read Full Article…)
Posted on 29 December 2010
I was thinking the about computer system recovery, mostly because I was/am trying to fix a computer at home. It started me thinking about the companies I have seen over the years with the blank check approach to DR and the ones I have seen with the piggy bank approach. It’s interesting, I’ve seen the companies that wanted/needed 100% uptime, but they are in denial of the costs associated to that level of uptime. Anyways, I was wondering how the cloud (a service provider) might be a solution for some types of DR for those companies with little to no budgets for IT.
Let’s use a small company as the example, they don’t have data centers, they have a room with ‘servers’ in it. A file server, email server, company web server, database server, etc. They have some spare parts in the closet, but not everything. The servers are under warranty, so replacements are available but it takes some time to get the parts ordered and shipped. The company isn’t in a situation that they are ready to deal with the costs associated to seamless DR, but they do want limited downtime.
If you think about it, if you had a server that was the company web server, email server or a database of some sort, and you had some type of hardware failure like the motherboard getting fried (and no spare available). If you have good documented procedures for restoration of the server, you could hit up a cloud provider to get a OS provisioned and then at least set up a temporary “server” while the other server (or parts) are being ordered/fixed. If you have some level of virtualization adoption, this type of DR might be even easier. The point is, if you are on a low budget, you could test out these type of DR scenario’s up front and then not have to have the hardware costs associated to DR.
With the amount of providers out there that are making it easier and easier to click a few times and then have an OS provisioned and available quickly, using the cloud as an alternative DR is becoming more of an option. There are some providers that are making canned configurations available which will make it even easier. Anyways, I have to go, I need to get back to fixing my computer problem, the CEO (wife) is complaining about how the IT department (me) is not responsive enough to fixing the outage. She has important web surfing to do 🙂