Blog

As a composer, there are two major aspects of my job that I need to be 100% impeccable with. The first is creating emotion and atmosphere suitable for each individual project I work on. The second, of equal importance, is content delivery. 

Each project I work on has its own needs. Potential clients can view my work on my website, and I may send them a demo piece that will specifically cater to their project on an atmospheric and emotional level. After my collaborators and I have settled on the feel of the music for a project, it’s off to production of my work. 

I draw inspiration from many sources- movies, television shows, video games, and sometimes more literary materials. I internalize that inspiration and begin working. I often use the “color palette” created in my demo piece for the whole of the project to create a sense of consistency and a signature sound for the project. When all creative hurdles are overcome, I can tie a neat little bow on my part of the project and call it done. 

While I use various websites in each phase of a project to get clients my work, SSD Nodes is the first stop for any of my clients. By extension, it is also the most important. SSD Nodes, from day one, has provided the best service possible for a cloud host- and it shows! 

With SSD Nodes, potential clients don’t have to wait long for my website. They can check out my work almost literally in the blink of an eye. I’ve even gotten comments from website visitors about how fast and reliable my website is. And as I am still in the infancy of my career as a composer, that’s remarkably helpful, whether it is noticed or taken for granted. 

I first got on board with SSD Nodes in summer 2011, which is also when I started to see myself getting quite a bit of work as a composer. SSD Nodes’ establishment could not have been more impeccable. I wish to thank SSD Nodes and their employees for all their hard work they do everyday, and for how they have propelled my future forward. I’m glad to be their client. 

Isaac Vail / http://isaacvail.com / http://www.imdb.com/name/nm4455071/

We’re proud to welcome James Kim to our team as the new A/R Representative. A/R stands for “Accounts Receivable”, a “meta”-position created for someone to maintain and improve client relations, whether it be through helping a prospective client find the right package for their needs, assisting with billing, handling most back-end accounting/reporting and coordinating with our managed services clients.

We brought James on board a few months ago and he worked diligently behind the scenes and exceeded our expectations by not just fulfilling the position requirements, but using his diverse knowledge of networking and systems administration to bring our products and support to a new level.

We have a lot of new and exciting features coming out within the next few months, and these would not have been possible without the invaluable addition of James to our team.

-Matt

The price-per-gigabyte of SSDs doesn’t always fit the budget when planing your future projects. This post will help you decide when an SSD is an incredible asset and when it isn’t (although it will still try convince you that an SSD volume is useful anyway!).

Micron (the company that makes our SSDs) did a very interesting video comparing an array of 15k SAS drives versus the same model SSDs we use: http://www.youtube.com/watch?v=og73pkYVKRk

Our SSDs provide substantially less latency than typical rotational drives, which results in much faster accesses and less time the CPU has to spend waiting for the data before processing it. The drives we use offer up to 45,000 Random Write IOPS (Input/Output operations Per Second) and up to 360MB/s reads, whereas most SATA rotational drives offer only a few hundred random write IOPS and up to 120MB/s reads. With this price point though, SSDs are more expensive per gigabyte than SATA, but still incredibly cheaper than RAM.

There are three main workloads: heavy reads, heavy writes, and equal reads/writes. All three can benefit from the use of SSDs, either using the volume directly or as a cache for your rotational drives.

Heavy reads can see improved performance by read caching, as all you’ll need is a sufficient amount of RAM and a caching engine (varnish, memcached, etc) between the user and database, or frequently accessed data on your rotational drives.

Heavy writes can see improved performance with either putting the data directly on an SSD volume or using an SSD volume as a write cache. We also have a few cases of clients using their SSD volumes as write caches for ZFS (we helped set it up), and they seem happy (they continue paying for it). Their workload is unique, so not many people would need a solution like that.

Freaky combination of both? Consider utilizing ZFS and an SSD volume as a cache. It’s cheaper than RAM and extremely scalable.

Take a look at your particular workload with a simple tool like “ioping” (http://code.google.com/p/ioping) and look at the latency of your I/O operations, and use “iostat” to get an idea of how many transactions per second (tps) your drives can handle. This will help you determine if an SSD is right for you.

-Matt

A provider should NEVER give the false sense that your data is backed up when it isn’t. It also falls upon everyone to have an offsite, backup and archival location that’s completely separate from your primary provider. Even if a provider is backing up your data, you should still have an offsite location (even if it’s a desktop server at home that runs rsync backups daily). Think about it: if your site is losing thousands of dollars per day when it’s not online, investing a few hundred in hardware, drives and a business internet connection should be a no-brainer.

When Amazon’s EC2 back-end had issues on April 22, 2011 there were many individuals and businesses that were unable to process anything on their EBS volumes. Not surprisingly, the only people affected were the ones who didn’t architect a proper failover and disaster recovery plan.

One particular instance that sticks in my memory was someone claiming to be “monitoring hundreds of cardiac patients at home”.

Original Forum Post - https://forums.aws.amazon.com/thread.jspa?threadID=65649

“We are a monitoring company and are monitoring hundreds of cardiac patients at home. We were unable to see their ECG signals since 21st of April”

“Not restored. Not heard from Amazon. People out there - please take a look at our volumes! This not just some social network website issue, but a serious threat to peoples lives!”

Avoiding These Problems

The best way to avoid this is to architect your backups and infrastructure among multiple providers to rule out any single point of failure. Keep them up to date, and check them often!

Practice recovery and failover processes so when the time comes you’re ready and trained to handle it. Always keep your staff and clients informed.

-Matt

I get asked pretty often what I do for a living (there’s just no way to get around the question). For the majority of people, just mentioning “I work with computers” is enough of an answer to steer away the conversation. There are two groups of people that will ask me follow up questions: 1) the people who already work in IT, and 2) the people who are genuinely interested, but have no clue about the subject. I enjoy talking to both groups equally, but I’m going to focus most of this post on the second group.

The real truth to myself of knowing whether or not I understand something is to be able to discuss it with someone who has never had an introduction to the field. If I’m able to clearly convey the topic, then I’ve just reaffirmed to myself that I understand the subject.

One example I’ll tell people in the second group is: let’s say you bought yourself a $5,000 Mac Pro (which is the entry price of the 12-core model). Now, let’s say you didn’t need to use it 24/7, but if you were doing moderate video editing 1-2 days per week. To offset your costs, you have a few friends who want to pitch in and use it at random times as well, but don’t necessarily need the full power or resources. So everybody pitches in $1,000 to use the capacity of a $5,000 Mac Pro.

At this point, most people nod with an “OH I see!” expression and remark about that being a cool idea. 

I was talking to my fiancee’s sister last week about this, explaining the Mac Pro analogy, and in the middle of it she asks me “Why would I ever need a node?” I was stumped for a while. My brain did a bubble sort on the information I knew about her, which resulted in throwing a 404 to the query. Slowly, I retorted “Because they’re cool”. I’m an engineer who masquerades as a salesman (our team is small and we all wear multiple hats), so I’ll never sell anything if the person doesn’t have a want, need or any use for, but I was still surprised.

A while later, I was thinking of all these random solutions for things she could do with it, like using it to back up all her photos, browsing the internet securely from public wifi areas, but I kept coming to the same conclusion that she just didn’t need it.

Then it hit me.

She’s running a medical billing company, she could use a node as a remote desktop for her work (she uses multiple computers, so having a standard desktop would be greatly beneficial)!

We don’t have two identical clients. More than half of our clients are running on some sort of custom package that fit their needs. It’s important to understand that most people don’t know what they want, or understand the technology and how it could make their lives easier. It’s our job to be a conduit of expression and ideas, and to be permeable enough where people can build their own platforms with our services.  

So, how would you describe virtual/cloud computer to someone who isn’t that familiar with computers?

-Matt

I get the occasional request here and there about adding a large number of IPv4 addresses routed to a single node. It becomes increasingly difficult as our allocations are already tight, and services (for the most part) don’t need to be staggered among multiple IPs.

Take for instance, you’re running mail, dns, and http/https services on one of our nodes. Your main concern should be getting the reverse DNS entries to match the hostname of the mail server (otherwise your mail will be rejected by most providers), and getting your services configured properly. Your https (SSL) service doesn’t need multiple IPs, just make sure the virtual hosts are set up properly.

One of the main selling points of our nodes is providing full native IPv6 support.We’ve been handing out about 65,000 IPv6 addresses per node, which (we admit) is way more than you’ll need. The only kinds of people that would need that many addresses are either doing shady SEO or spamming.

So please, take a look at your infrastructure with this idea in mind: “Which services can I consolidate from 20+ IPs to one or two IPs?” and consider dual-stacking on both protocols.