The Amazon Outage

As of this morning (April 26), Amazon was still listing “Instance connectivity, latency and error rates” for a very small percentage of clients at its North Virginia center.  Thus the outage, which started on April 21 and affected a number of well-known companies, has lasted nearly 5 days. This is not the first Amazon outage: if you Google “Amazon outage” you will find entries for 2011, 2010 and 2009.  A Forbes blogger entitled it “The Day the Cloud Died” in a rendering of “American Pie.” It will not be the last.

What does it mean for SaaS products?  First, it must be said that the down time for major players such as Amazon or Rackspace is significantly less than for most medium size law firms that have their own servers in-house. Of course, this is no consolation when your provider does have an outage.  I think the difference is largely psychological: when an in-house system goes down, there is a sense that you can “do something” about it, and there is certainly someone to yell at.  Try yelling at Amazon. Doesn’t work so well.

As a practical matter, the issue is: what sort of backup system does your SaaS provider have in place?  So if there is an outage in one location, can they switch over to another location?  Do they offer, as NetDocuments does, a “backup” to in-house servers, so that if the Internet goes down, you can switch over to functioning locally? 

As Jack Newton of Clio notes in a White Paper on “The Ethics and Security of Cloud Computing,” “Geographic redundancy, or geo-redundancy, takes advantage of multiple, geographically distributed data centers. The impact of an outage at one data center can thus be minimized by automatic failover to additional data centers. ...  The SaaS provider should, at a minimum, be performing daily backups of all data and storing this backup in a secure, offsite location. Ideally, backups should be performed multiple times per day, and replicated to multiple, secure offsite locations.”

So moving to the cloud should still be on your list of options. You do, however, need to do due diligence about the provider’s security and backup options. You may find you have to pay more to get proper backup (as with NetDocuments), but like any type of insurance, it is probably worth it.

Comments

[Frank Rivera]

John,

It wasn't access to ones data that was the problem, it was access to the applications that were hosted.

This is the problem when someone that is suppose to be hosting your data isn't actually in control. MyCase (a legal practice management product in the Cloud) was down for 17hours. Who is the Service Provider here? Who did the law firms sign a terms of service agreement with? Amazon or MyCase?

What would the Judge say if you said "Your Honor, I was unable to file the Motion because Amazon was down yesterday"? Go ahead and hand him a study.

Right now the Mac I'm using has an uptime of nearly 14 months and the one to my right had an uptime of 9 months before I unplugged it to take it to the ABA TechShow in Chicago. So I don't buy it, no matter how many studies you toss at me not every system that isn't in the Cloud is less stable or has more outages, breeches, etc.

As I have said on numerous occasions. Just because everyone does something one way DOES NOT make it "best practice". It makes it "common practice". The two are totally different.

Solution: A Hybrid Single Tenancy (HST) architecture is the only architecture a law firm's data and mission critical apps should be placed onto. The vendor you sign up with should have absolute complete control over the devices you data and apps run on. If they can't walk up to a machine and point to the machine your data is sitting on then they are outsourcing your obligation to protect your client's data. This is not to say that a HST is less likely to have an outage but, it is less likely to have a complete outage. If a application instance or database was to go down, get confiscated by the feds, or attacked by hackers all the other clients in the system are un affected. Client's have their own database and instance of the application but still share resources. This is no pipe dream. We built this sort of fault tolerance into Time Maters Word server and that was back in 2002. You can't reach server one then try server two, can't reach that then spin up a new instance now 1000km away. Today, we have built a resilient system from the ground up designed specifically to accommodate law firm data in a HST and a legal practice management application that only requires a browser to run. In fact it is so resilient that it can be hosted in house in what we call a Private Cloud. Yes, another buzzword term that has gotten some press lately but, we have used it to describe HoudiniESQ installed On-premise for over two years. One code base, run anywhere.

So law firms wanting/needing remote access to their data now have a choice and they don't have to compromise much for remote access either as we lead the pack in features, integration, and performance. If they aren't comfortable with the Cloud just yet, no problem, if they want to move up to the Cloud at a later time or back down to a On-premise install no problem. Having a Choice is pretty nice.

THIS is "best practice". Just FYI from a Legal Services Providers perspective. NOT ALL SaaS Providers are the SAME.

Frank Rivera CEO HoudiniESQ

[Frank Rivera]

an another thing.

It is easy to point out that only a small percentage of SaaS vendors were affected by Amazon's outage (hundreds in fact). That is no comfort since a small number of those vendors actually lost data as reported by Amazon.

How much data loss is acceptable? To many moving parts means more can and will go wrong.

In the end what will make all the difference in the world of Cloud based apps and data storage is who is actually in control. Frankly, if you store a Law Firms data on servers you do NOT control then you are obligated to inform every Law Firm you do business with of this fact up front.

Frank Rivera CEO HoudiniESQ

[Ecommerce website developers]

I definitely agree with your points that So moving to the cloud should still be on list of options. however, need to do due diligence about the provider’s security and backup options. You may find you have to pay more to get proper backup (as with NetDocuments), but like any type of insurance, it is probably worth it.