nethergoat

Bizo (http://bizo.com) is hiring a Web/UI/UX Designer in downtown San Francisco. Essentially, we need someone to own design and usability for everything we do. Full description here: http://bit.ly/adbvhQ

Bizo is a year and a half old and already profitable. In December, we closed a $6 million round from Bessemer Venture Partners and placed second in the Amazon Web Services Startup Challenge. We have ~17 people (7 engineers) across the country, and we'll be moving our SF headquarters into a new space (Commercial & Sansome) next week.

If interested, drop me a line: mike@bizo.com

How is this any different from a large corporation? Having been on both sides, I'd argue that Amazon's mean time to response/resolution is as good as or better than what is generally seen in silo'd enterprise shops (provided you pay for AWS support- strongly recommended for those with mission-critical cloud infrastructure).

How much traffic was each LB seeing? The use of round-robin DNS does not always result in an even distribution of traffic.

Deepak Singh of Amazon Web Services maintains a great list of (cloud-focused) parallel computing frameworks and platforms: http://deepaksingh.net/Resources/Computing_in_the_Cloud

He's on twitter, too: http://twitter.com/mndoci

I just joined the Bizo team three weeks ago. Great company, fantastic people (~15 total now), and, as Donnie said, we're growing very quickly.

Bonus points for foosball, ping pong, or GH/RB skills.

Note that this feature is (for now, at least) designed to add weight to content _created_ by your social circle. It does not, as I first thought, modify your search results based on friends' preferences.

From Google's FAQ: "Google Social Search is a feature designed to help you discover relevant publicly-accessible content from your social circle, a set of online friends and contacts. The idea is that content from your friends and social contacts is often more relevant to you than content from strangers. For example, a movie review from an expert is useful, but a movie review from your best friend can be even better."

More info here: http://www.google.com/support/websearch/bin/answer.py?hl=en&...

Yes, just over 1.5 years at ShareThis. ~150 nodes managed by RightScale for ~$4000/mo (Premium account plus an add-on account for developers). In this case, the "managing" includes rich metadata, server provisioning, configuration management, deployment scripts, monitoring, alerting, etc. You can basically think of it as [Chef/Puppet/CFEngine]+Nagios+Cacti+AWS console+misc other tools, all wrapped up in a pretty (and very functional) GUI.

CloudKick's pricing puts it at about half that of RightScale (at our scale - I'm not sure how RightScale is priced at the bottom end), but it looks like the features are monitoring/alerting focused - the management tools appear to be largely undeveloped.

CloudKick is certainly making good headway feature-wise, and I'm glad to see them spin up their revenue stream. I look forward to seeing what other features they add on in the future.

Per the comments below the article, you can disable this by disabling the 'send anonymous usage statistics' option.

Another commenter addressed the point about 1TB drives. I'd like to help on the bandwidth side.

A T1 is simply not adequate for a web company for two reasons: 1) it is a high-impact single point of failure (any issue here results in 100% outage); 2) it is limited to 1.5mpbs. I guarantee you Heroku is beyond 1.5Mbps at peak.

To make this a fair comparison, you need to look at burstable bandwidth. Unless you want to manage your own redundant BGP routers and multiple carrier lines, you'll want blended bandwidth provided by your datacenter or a specialized company (InterNAP, etc). This bandwidth will run you $50-100/Mbps/mo (let's say $75) and is billed at 95th percentile of usage. A year and a half ago when my company's EC2 network transfer costs were ~$8,000/mo (with a total EC2 bill of ~$30,000/mo), I worked out a bandwidth need of ~60Mbps (at 95th%) - this would represent a bandwidth cost of $4500/mo. Let's assume Heroku's EC2 cost breakdown is similar, and that their needs are 2/3 of what ours were then - we're left with $3,000/mo for bandwidth.

Yes, you're still paying a premium, but the cost savings of colo hosting are quickly offset by the additional manpower required to maintain physical servers. Whatever difference still remains is easily justifiable given the amount of flexibility the cloud affords you. This is exponentially more important for a startup where agility can make or break you.

You're right, that's not all they have. 22 is the number of m2.2xlarge nodes they run. There's no mention of how many other nodes they have running.

That being said, given how many toy, no-traffic apps are likely running (for free!) on the platform, I think we can safely assume there's a very high degree of multi-tenancy.

Agreed, that's a whole lot of eggs in one basket. I'd be interested to hear their reason for not spreading across multiple availability zones. My first thought was that m2.2xlarge nodes, since they're relatively new, might still be restricted to a single AZ; on further investigation, however, this does not appear to be the case.

As someone who has spent the last year and a half running a 200-(persistent)node environment on EC2, including multiple m1.large and m1.xlarge DB pools, I can assure you those stories stem from FUD and unreasonable expectations.

Yes, EBS is not very fast, especially compared to local disk. You can work around this, however, by configuring multiple volumes in a RAID configuration (as you have mentioned), or by scaling out with additional nodes. The size and workload of your database will dictate which is more cost-effective.

Spooling up many nodes to find the "best" one is completely unnecessary. In my experience, EC2 nodes have been remarkably consistent in performance. I won't say I run a load test on every one, but I will say that over 100,000 node launches, I've never had to shut down a poorly-performing instance that couldn't be attributed to a hardware issue (rare, and for which Amazon sends notifications).

Don't listen to the naysayers. Come on in, the water's fine!

I'm not a member, but this looks promising: http://www.linkedin.com/groups?gid=86116&trk=anetsrch_na...

The big guns like Google, Yahoo, and Microsoft are sure bets. Social networks like Facebook and LinkedIn also have large ML teams and probably have decent internship programs.

Many startups also have similar interships - try http://ventureloop.com

Given that S3 must be used as the origin, I suspect live streaming is not possible at this time (S3 does not support streaming upload).

http://highscalability.com/ is a great resource for building scalable sites - it has consolidated architectural details (culled from presentations, blog posts, interviews, and more) for many large-scale sites (YouTube, Facebook, Amazon, etc).

As for testing environments, take a look into the principles of continuous integration (many resources available) and continuous deployment (especially Eric Ries's blog: http://www.startuplessonslearned.com/). Some tools commonly used to this effect: Git/SVN for source control; Hudson/Cruise Control/Team City for builds; Selenium for testing.

Finally, applying patches and updates to a server falls into the domain of systems administration, so I'd suggest looking for articles on patching/updating high-availability sites. Note, however, that cloud computing is a huge game-changer here. Whereas at many enterprises, applying patches is an arduous (and tedious) quarterly effort culminating in an late-night maintenance window with many hands on deck (all crossing their fingers), clouds afford their adopters the ability to simply fire up new, updated servers alongside the old ones, allowing them to be thoroughly tested in place before any traffic is redirected.

-Mike (@mikebabineau)

This is tremendously useful and lowers the barrier to entry for new adopters. Unfortunately, it's also a temptation to make poor design decisions.

One of the key benefits of EC2 and other cloud offerings is that, in a properly automated environment, it is faster and easier to replace a faulty node than to troubleshoot it. The same is true for upgrades - simply boot a new instance from an updated AMI or Chef/Puppet/Cfengine/RightScale config and point traffic at the new server. Minimal wheelspin, minimal downtime.

Those who use this new feature to take a, for lack of a better term, "server-hugger" approach to cloud computing will miss out on this. Though it may reduce their cloud onboarding time, it's inefficient in the long term and will be an operational burden as the environment grows.

The author, Pete Skomoroch (http://twitter.com/peteskomoroch), will be giving a talk about TrendingTopics.org at Hadoop World NYC this Friday. Registration is still open ($400): http://www.cloudera.com/hadoop-world-nyc

My company (startup, ~25 employees) also used the free version for several months (late '08 and early '09) but were disappointed with the sound quality (we, too, never tried the paid version).

We used GoToMeeting for several months as well, but got fed up with frequent issues (dropped calls, failures to connect to join meetings, etc. almost daily).

We now use WebEx. We've had very few issues and the quality is good, but the interface is a real pain in the ass to use.

I'd be interested in hearing anyone's experience going from the free version to the paid one.

"Today’s incident also shows the fragility of the 'cloud' as it can be knocked out a single lightening strike."

How is this different from any other data center? Note too that on EC2, your instances will be spread randomly throughout the facility - you'd actually be better off having this happen on EC2 vs. in a traditional DC. Furthermore, this was a localized incident inside of a single availability zone - the other four (now five, actually) AZs were completely unaffected.

The author clearly does not know what he is talking about.

I'm glad one commenter on the article had the sense to call him out. Too bad he was followed by the usual set of trolls.

Dolores Labs has great blog posts on some data mining and visualization experiments, definitely worth checking out: http://blog.doloreslabs.com/

Glad to see they're hosting a MTurk meetup. I'd love to attend, but it conflicts with another event.

We have several application components that use software load balancers (HAProxy) so we can manually scale each component quickly. The load balancing service will let us shut down each of those instances for a decent savings. Our QA environment is a copy of prod (scaled-back, but identical, including LBs) - more nodes to shut down, more money saved.

This is easy to implement and is basically found money, so we'll be doing it ASAP.

My expectation is that an Amazon (LB) access point is more reliable than a generic EC2 instance. For these components, we'll be paying less for something that's both more reliable (hopefully) and easier to manage.

Auto-scaling is powerful and easy to set up (even in the private beta, before tools were available), but as we're heavily integrated with RightScale (specifically, using templates throughout) migrating to pre-configured AMIs would take some effort. We'll have to do some internal review before making a decision on rolling out auto-scaling to existing apps.

Monitoring is, as expected, also provided by RightScale. Amazon requires you have CloudWatch turned on for instances in auto-scaled pools (for obvious reasons), so that's one way we may end up using it. RightScale's blog post mentioned they'll be integrating their alerting and reporting with CloudWatch, but I'm not sure it's worth the extra $0.015/intance-hour to have the data come from Amazon instead of RightScale. I expect that, for now, CloudWatch is more relevant to: 1) users not on RightScale, Cloudkick, etc. 2) users who want direct access to their monitoring data (for archival, custom graphs, whatever - not currently possible with RightScale, not sure about other cloud management vendors)

Amazon's move up the application stack (Elastic MapReduce, for starters) may create additional use cases for CloudWatch, but that's entirely up to Amazon.

What I'd really like to see in the near-term is an open source tool to read in and graph CloudWatch data that's been archived somewhere (SimpleDB would be cool, but S3 would work too). Of course, I can see why people would be hesitant to start down that path given that Amazon has a console that's begging to include this functionality.

-Mike B @ ShareThis

For those already in decent shape, here's a way to maximizing the use of those 20 minutes: high-intensity interval training (HIIT - http://en.wikipedia.org/wiki/High-intensity_interval_trainin...). The basic premise is to alternate between maximum effort (sprinting) and medium effort (jogging) at set intervals. The key is that the sprints are anaerobic rather than aerobic. This type of workout induces a state of excess post-exercise oxygen consumption, thereby raising your resting metabolic rate for about 24 hours.

Hop on a treadmill (or bike, elliptical, whatever), warm up for 5 minutes, then go all-out for 30 seconds. Jog for 60 seconds, then go all-out for another 30 seconds. Repeat the cycle 6-8 times. Cool down for 5 minutes.

Best bang for your 20 minutes. Do it just three times a week; add in lifting, normal cardio, etc. as desired.