Tag Archives: EBS

Is AWS the patient that needs constant medication?

storm coming

I was just reading High Scalability about Why Swiftype moved off Amazon EC2 to Softlayer and saw great wins!

We’ve all heard by now how awesome the cloud is. Spinup infrastructure instantly. Just add water! No up front costs! Autoscale to meet seasonal application demands!

But less well known or even understood by most engineering teams are the seasonal weather patterns of the cloud environment itself!

Join 28,000 others and follow Sean Hull on twitter @hullsean.

Sure there are firms like Netflix, who have turned the fickle cloud into one of virtues & reliability. But most of the firms I work with everyday, have moved to Amazon as though it’s regular bare-metal. And encountered some real problems in the process.

1. Everyday hardware outages

Many of the firms I’ve seen hosted on AWS don’t realize the servers fail so often. Amazon actually choosing cheap commodity components as a cost-savings measure. The assumption is, resilience should be built into your infrastructure using devops practices & automation tools like Chef & Puppet.

The sad reality is most firms provision the usual way, through the dashboard, with no safety net.

Also: Is your cloud speeding for a scalability cliff

2. Ongoing network problems

Network latency is a big problem on Amazon. And it will affect you more. One reason is you’re most likely sitting on EBS as your storage. EBS? That’s elastic block storage, it’s Amazon’s NAS solution. Your little cheapo instance has to cross the network to get to storage. That *WILL* affect your performance.

If you’re not already doing so, please start using their most important & easily missed performance feature – provisioned IOPS.

Related: The chaos theory of cloud scalability

3. Hard to be as resilient as netflix

We’ve by now heard of firms such as Netflix building their Chaos Monkey to actively knock out servers, in effort to test their ability to self-healing infrastructure.

From what I’m seeing at startups, most have a bit of devops in place, a bit of automation, such as autoscaling around the webservers. But little in terms of cross-region deployments. What’s more their database tier is protected only by multi-az or just a read-replica or two. These are fine for what they are, but will require real intervention when (not if) the server fails.

I recommend building a browse-only mode for your application, to eliminate downtime in these cases.

Read: 8 questions to ask an aws expert

4. Provisioning isn’t your only problem

But the cloud gives me instant infrastructure. I can spinup servers & configure components through an API! Yes this is a major benefit of the cloud, compared to 1-2 hours in traditional environments like Softlayer or Rackspace. But you can also compare that with an outage every couple of years! Amazon’s hardware may fail a couple times a hear, more if you’re unlucky.

Meanwhile you’re going to deal with season weather problems *INSIDE* your datacenter. Think of these as swarms of customers invading your servers, like a DDOS attack, but self-inflicted.

Amazon is like a weak immune system attacking itself all the time, requiring constant medication to keep the host alive!

Also: 5 Things toxic to scalability

5. RDS is going to bite you

Besides all these other problems, I’m seeing more customers build their applications on the managed database solution MySQL RDS. I’ve found RDS terribly hard to manage. It introduces downtime at every turn, where standard MySQL would incur none.

In my experience Upgrading RDS is like a shit-storm that will not end!

Also: Does open source enable the cloud?

Get more. Grab our exclusive monthly Scalable Startups. We share tips and special content. Our latest Why I don’t work with recruiters

If you use MySQL in the Amazon cloud, you need to ask yourself this question

Join 25,000 others and follow Sean Hull on twitter @hullsean.

Are you serious about backups?

If you’re just using Amazon EBS snapshots, that may not be sufficient. There’s a good chance it won’t protect you against your next data loss.

That’s why I like to have a few different types of backups

Also: 5 more things deadly to scalability

Protect against operator error

mysqldump is a tool every DBA is familiar with. Same as a hotbackup or snapshot you say? Just more labor? Not true.

A dump allows you to restore one table, or one schema. That’s why they’re also known as logical backups. What’s more you can edit the file, remove indexes, change object names, or datatypes. All these can be essential in the screwy and unpredictable event of a real world outage.

Expect the unexpected!

Read: Why devops talent is in short supply

Test those backups regularly

If you haven’t actually tried to restore, you really don’t know if you have everything. Did you backup stored procedures & database code? How about grants? Database events? How about cronjobs? What about the my.cnf file? And your replication configuration?

Yes there are a lot of little pieces, and testing your backups by rebuilding everything is an attempt to poke holes in your plan, and hit issues before d-day!

Related: MySQL interview guide for managers and candidates alike

Replication isn’t a backup

Replication is getting better and better in MySQL. It used to fail regularly. MyiSAM was very unpredictable. But even in the comfortable realm of Innodb, there can still be data drift. If you’re on MySQL 5.0 or 5.1, you should consider performing regular checksums. These test the integrity of data and compare what’s actually in master & slave. Bulletproofing MySQL replication with checksums.

Read: Why high availability is so very hard to deliver

Have you considered security around your backup files?

While you’re thinking about backups, make sure the files themselves are secure. Remember they contain your crown jewels. Hopefully individual data that’s sensitive is encrypted, but still you should secure their final resting place as well.

If you’re using S3, consider encrypting the file before shipping it up to the bucket.

Read this: Why a four letter word divides dev and ops

Get more. Grab our exclusive monthly Scalable Startups. We share tips and special content. Our latest Why I don’t work with recruiters

5 cloud ideas that aren’t actually true

storm coming

Join 20,000 others and follow Sean Hull’s scalability, startup & innovation content on twitter @hullsean.

Cloud computing is heralding us into a wonderful era where computing can be bought in small increments, like a utility. This changes the whole way we plan, manage budgets, and accelerates startups making them more agile.

But it’s not all wine & roses up there. I’ve heard a few refrains from clients over the years, and thought I’d share some of the most common.

1. Scaling is automatic

Rather recently I was working with a client on building some sophisticated reports. They needed to slice & dice customer data, over various time series, and summarize with invoices & tracking data. Unfortunately their dataset was large, in the half terabyte range.

Client: Can we just load all this data into the cloud?
Me: Yes we can do that. Build a system in Amazon public cloud, can support large datasets.
Client: I want it to scale easily. So we won’t have these slow reports. And as we add data, it’ll just manage it easily for us.
Me: Well it’s a little bit more complicated than that, unfortunately.

Unfortunately this is a rather familiar conversation that I have quite often. A lot of the press around cloud scalability, centers around auto-scaling, Amazon’s renowned & superb virtualization feature. Yes it’s true you can roll out webservers to scale out this way, but that’s not the end of the story. Typically web applications have a lot of components, from caching servers, to search servers, and of course their backend datastore.

But can we scrap our relational database, such as MySQL and go with one that scales out of the box like Riak, Cassandra or Dynamodb?

Those NoSQL solutions are built to be distributed from the start, it’s true. And they lend themselves to that type of architecture. However, if you’ve built up a dataset in MySQL or Oracle, and more so an application around that, you’ll have to migrate data into the NoSQL solution. That process will take some time.

Like teaching a fish to fly, it make take some time. They do well in water, but evolution takes a bit longer.

Related: RDS or MySQL 10 use cases

2. Disaster recovery is free

In the traditional datacenter, when you want DR, you setup a parallel environment. Hopefully not in the same room, same city or same coast even. Preferrably you do so in a different region. What you can’t get around is dishing out cash for that second datacenter. You need the servers, just in case.

In the cloud, things are different. That’s why we’re here, right? In amazon you have regions already setup & available for plugin-n-play use. Setup your various components, servers, software & configure. Once you’ve verified you can failover to the parallel environment you can just turn off all those instances. Great, no big charges for all that iron that you’d pay for to keep the rooms warm in an old-school datacenter. Or do you?

As it turns out, since you don’t have this environment running all the time, you’ll want to test it more often, run fire drills to bring the servers back online. That’ll incur some costs in terms of manpower. You’ll also want to include in there some scripts to start those servers up, and/or some detailed documentation on how to do that. And don’t lose that documentation, either will you?

You may also want to build some infrastructure as code unit tests. Things change, code checkouts evolve, especially in the agile & continuous integration world. Devops beware!

Read this: Why a killer title can make or break your content efforts

3. Machines are fast

Fast, fast, fast. That’s what we expect, things keep getting faster, right? Hard to believe then that the world of computing took a big step backward when it jumped into the cloud. Something similar happened when we jumped to commodity Linux a decade ago.

In amazon, it’s a multi-tenant world. And just like apartment buildings, popular restaurants, or busy highways you must share. When things are quiet you may have the road to yourself, but it’ll never be as quiet as a dirt road in the country!

Amazon is making big strides though. They now offer memory optimized & storage optimized instances. And an even bigger development is the addition of the most important feature for performance & scalability. That said the network & EBS can still be a real bottleneck.

Also: What is a relational database & why is it important?

4. Backups aren’t necessary

I’ve experienced a few horror stories over the years. I wrote about one noteworthy one When fat fingers take down your business.

True EBS snapshots make backing up your whole server, well a snap! That said a few extra steps have to happen (flush the filesystem & lock tables) to make this work for a relational database like MySQL or Oracle. And suddenly you have a verification step that you also need to perform. You see no backups are valid until they’ve been restored, remember?

But even with these wonderful disk snapshots, you’ll still want to do database dumps, and perhaps table dumps. Operator error, deleting the wrong data, or dropping the wrong tables, will always be a risk. Ignore backups at your own peril!

Check this: Why CTOs underestimate operational costs

5. Outages won’t happen

In an ideal world, everything is redundant, and outages will be a thing of the past. We’ll finally reach five nines uptime and devops everywhere will be out of work. :)

It’s true that Amazon provides all the components to build redundancy into your architecture, and very cutting edge firms that have taken netflix’s approach with chaos monkey are seeing big improvements here. But AirBNB did fail and at root it was an Amazon outage that shouldn’t ever happen.

Read: Why Oracle won’t kill MySQL

Get more. Monthly insights about scalability, startups & innovation.. Our latest Are SQL Databases Dead?

Review: Host Your Web Site In The Cloud, Amazon Web Services Made Easy

Jeff Barr’s book on AWS is a very readable howto and a quick way to get started with EC2, S3, CloudFront, CloudWatch and SimpleDB.  It is short on theory, but long on all the details of really getting your hands dirty.  Learn how to:

  • get started using the APIs to spinup servers
  • create a load balancer
  • add and remove application servers
  • build custom AMIs
  • create EBS volumes, attach them to your instances & format them
  • snapshot EBS volumes
  • use RAID with EBS
  • setup CloudWatch to monitor your instances
  • setup triggers with CloudWatch to enable AutoScaling

I would have liked to see examples in Chef rather than PHP, but hey you can’t have everything!

Review: Host Your Web Site In The Cloud by Jeff Barr