Yesterday Amazon's S3 service had an outage that lasted about six hours. Unsurprisingly this has led to a bunch of wailing and gnashing of teeth from the very same pundits that were hyping the service a year ago. The first person to proclaim the sky is falling is Richard MacManus in his More Amazon S3 Downtime: How Much is Too Much? who writes

Today's big news is that Amazon's S3 online storage service has experienced significant downtime. Allen Stern, who hosts his blog's images on S3, reported that the downtime lasted 3.5 over 6 hours. Startups that use S3 for their storage, such as SmugMug, have also reported problems. Back in February this same thing happened. At the time RWW feature writer Alex Iskold defended Amazon, in a must-read analysis entitled Reaching for the Sky Through The Compute Clouds. But it does make us ask questions such as: why can't we get 99% uptime? Or: isn't this what an SLA is for?

Om Malik joins in on the fun with his post S3 Outage Highlights Fragility of Web Services which contains the following

Amazon’s S3 cloud storage service went offline this morning for an extended period of time — the second big outage at the service this year. In February, Amazon suffered a major outage that knocked many of its customers offline.

It was no different this time around. I first learned about today’s outage when avatars and photos (stored on S3) used by Twinkle, a Twitter-client for iPhone, vanished.
…
That said, the outage shows that cloud computing still has a long road ahead when it comes to reliability. NASDAQ, Activision, Business Objects and Hasbro are some of the large companies using Amazon’s S3 Web Services. But even as cloud computing starts to gain traction with companies like these and most of our business and communication activities are shifting online, web services are still fragile, in part because we are still using technologies built for a much less strenuous web.

Even though the pundits are trying to raise a stink, the people who should be most concerned about this are Amazon S3's customers. Counter to Richard MacManus's claim, not only is there a Service Level Agreement (SLA) for Amazon S3, it promises 99.9% uptime or you get a partial refund. 6 hours of downtime sounds like a lot until you realize that 99% uptime is 8 hours of downtime a month and over three and a half days of downtime a year. Amazon S3 is definitely doing a lot better than that.

The only question that matters is whether Amazon's customers can get better service elsewhere at the prices Amazon charges. If they can't, then this is an acceptable loss which is already covered by their SLA. 99.9% uptime still means over eight hours of downtime a year. And if they can, it will put competitive pressure on Amazon to do a better job of managing their network or lower their prices.

This is one place where market forces will rectify things or we will reach a healthy equilibrium. Network computing is inherently and no amount of outraged posts by pundits will ever change that. Amazon is doing a better job than most of its customers can do on their own for cheaper than they could ever do on their own. Let's not forget that in the rush to gloat about Amazon's down time.

Now Playing: 2Pac - Life Goes On

 

For the past few years, the technology press has been eulogizing desktop and server-based software while proclaiming that the era of Software as a Service (SaaS) is now upon us. According to the lessons of the Innovator's Dilemma the cheaper and more flexible SaaS solutions will eventually replace traditional installed software and the current crop of software vendors will turn out to be dinosaurs in a world that belongs to the warm blooded mammals who have conquered cloud based services.

So it seems the answer is obvious, software vendors should rush to provide Web-based services and extricate themselves from their "legacy" shrinkwrapped software business before it is too late. What could possibly go wrong with this plan? 

Sarah Lacy wrote an informative article for Business Week about the problems facing software vendors who have rushed into the world of SaaS. The Business Week article is entitled On-Demand Computing: A Brutal Slog and contains the following excerpt

On-demand represented a welcome break from the traditional way of doing things in the 1990s, when swaggering, elephant hunter-style salesmen would drive up in their gleaming BMWs to close massive orders in the waning days of the quarter. It was a time when representatives of Oracle (ORCL), Siebel, Sybase (SY), PeopleSoft, BEA Systems, or SAP (SAP) would extol the latest enterprise software revolution, be it for management of inventory, supply chain, customer relationships, or some other area of business. Then there were the billions of dollars spent on consultants to make it all work together—you couldn't just rip everything out and start over if it didn't. There was too much invested already, and chances are the alternatives weren't much better.

Funny thing about the Web, though. It's just as good at displacing revenue as it is in generating sources of it. Just ask the music industry or, ahem, print media. Think Robin Hood, taking riches from the elite and distributing them to everyone else, including the customers who get to keep more of their money and the upstarts that can more easily build competing alternatives.

But are these upstarts viable? On-demand software has turned out to be a brutal slog. Software sold "as a service" over the Web doesn't sell itself, even when it's cheaper and actually works. Each sale closed by these new Web-based software companies has a much smaller price tag. And vendors are continually tweaking their software, fixing bugs, and pushing out incremental improvements. Great news for the user, but the software makers miss out on the once-lucrative massive upgrade every few years and seemingly endless maintenance fees for supporting old versions of the software.
…
Nowhere was this more clear than on Oracle's most recent earnings call (BusinessWeek.com, 6/26/08). Why isn't Oracle a bigger player in on-demand software? It doesn't want to be, Ellison told the analysts and investors. "We've been in this business 10 years, and we've only now turned a profit," he said. "The last thing we want to do is have a very large business that's not profitable and drags our margins down." No, Ellison would rather enjoy the bounty of an acquisition spree that handed Oracle a bevy of software companies, hordes of customers, and associated maintenance fees that trickle straight to the bottom line.
…
SAP isn't having much more luck with Business by Design, its foray into the on-demand world, I'm told. SAP said for years it would never get into the on-demand game. Then when it sensed a potential threat from NetSuite, SAP decided to embrace on-demand. Results have been less than stellar so far. "SAP thought customers would go to a Web site, configure it themselves, and found the first hundred or so implementations required a lot of time and a lot of tremendous costs," Richardson says. "Small businesses are calling for support, calling SAP because they don't have IT departments. SAP is spending a lot of resources to configure and troubleshoot the problem."

In some ways, SaaS vendors have been misled by the consumer Web and have failed to realize that they still need to spend money on sales and support when servicing business customers. Just because Google doesn't advertise it's search features and Yahoo! Mail doesn't seem to have a huge support staff that hand holds customers as it uses their product doesn't mean that SaaS vendors can expect to cut their sales and support calls. The dynamics of running a free, advertising based service aimed at end users is completely different from running a service where you expect to charge business tens of thousands to hundreds of thousands to use your product.

In traditional business software development you have three major cycles with their own attendant costs; you have to write the software, you have to market the software and then you have to support the software. Becoming a SaaS vendor does not eliminate any of these costs. Instead it adds new costs and complexities such as managing data centers and worrying about hackers. In addition, thanks to free advertising based consumer services and the fact that companies like Google that have subsidized their SaaS offerings using their monopoly profits in other areas, business customers expect Web-based software to be cheaper than its desktop or server-based alternatives. Talk about being stuck between a rock and a hard place as a vendor.

Finally, software vendors that have existing ecosystems of partners that benefit from supporting and enhancing their shrinkwrapped products also have to worry about where these partners fit in a SaaS world. For an example of the kinds of problems these vendors now face, below is an excerpt from a rant by Vladimer Mazek, a system administrator at ExchangeDefender, entitled Houston… we have a problem which he wrote after attending one of Microsoft's partner conferences

Lack of Partner Direction: By far the biggest disappointment of the show. All of Microsoft’s executives failed to clearly communicate the partnership benefits. That is why partners pack the keynotes, to find a way to partner up with Microsoft. If you want to gloat about how fabulous you are and talk about exciting commission schedules as a brand recommender and a sales agent you might want to go work for Mary Kay. This is the biggest quagmire for Microsoft – it’s competitors are more agile because they do not have to work with partners to go to market. Infrastructure solutions are easy enough to offer and both Google and Apple and Amazon are beating Microsoft to the market, with far simpler and less convoluted solutions. How can Microsoft compete with its partners in a solution ecosystem that doesn’t require partners to begin with?

This is another example of the kind of problems established software vendors will have to solve as they try to ride the Software as a Service wave instead of being flattened by it.  Truly successful SaaS vendors will eventually have to deliver platforms that can sustain a healthy partner ecosystems to succeed in the long term. We have seen this in the consumer space with the Facebook platform and in the enterprise space with SalesForce.com's AppExchange. Here is one area where the upstarts that don't have a preexisting shrinkwrap software businesses can turn a disadvantage (lack of an established partner ecosystem) into an advantage since it is easier to start from scratch than to retool.

The bottom line is that creating a Web-based version of a popular desktop or server-based product is just part of the battle if you plan to play in the enterprise space. You will have to deal with the sales and support that go with selling to businesses as well as all the other headaches of shipping "cloud based services" which don't exist in the shrinkwrap software world. After you get that figured out, you will want to consider how you can leverage various ISVs and startups to enhance the stickiness of your service and turn it into a platform before one of your competitor's does. 

I suspect we still have a few years before any of the above happens. In the meantime we will see lots more software companies complaining about the paradox of embracing the Web when it clearly cannibalizes their other revenue streams and is less lucrative than what they've been accustomed to seeing. Interesting times indeed.

Now Playing: Flobots - Handlebars

 

Sometime last week I learned that podcasting startup PodTech was acquired for less than $500,000. This is a rather ignominious exit for a startup that initially entered the public consciousness with its high profile hire of Robert Scoble and the intent to build a technology news media empire using RSS and podcasts instead of radio waves and news print.

When I first heard about PodTech via Robert Scoble's blog, it seemed like a bad business to jump into given the lessons of The Long Tail. The Web creates an overabundance of content and products, which is good for aggregators but bad for creators. Even in 2006 when PodTech was founded you could see this in the success of "Web 2.0" companies that acted as content aggregators like Google, YouTube, Wikipedia and Flickr while content creators like music labels and news papers were beginning to scramble for relevance and revenue. 

Kevin Kelly has a great post about this called Wagging the Long Tail of Love where he writes

So as one crosses the sections -- going from the short head to the long tail -- one should be consistent and view it from the aggregator's point of view or the creator's point of view. I think it is a mistake to conflate the two view points.

I've been wrestling with this for a while and I think the only advantage to the creator that I can see in the long tail is that aggregators can invent or produce a long tail domain that was not present before.  Like Seth's Squidoo does. Before Squidoo or Amazon or Netflix came along there was no market at all for many of the creations they now distribute. The proposition that long tail aggregators can offer to creators is profound, but simple: you have a choice between a itsy bitsy niche audience (with nano profits) or no audience at all. Before the LT was expanded your masterpiece on breeding salt water aquarium fishes from the Red Sea would have no paying fans. Now you have maybe 100.

One hundred readers/watchers/listeners is not economical. There is no business equation that can sustain profits for continual creation from so few buyers. (It can of course support the business of aggregation above the level of creation.) But the long tail niche creation operates perfectly well in the realm of passion, enthusiasm, obsession, curiosity, peerage, love, and the gift economy.  In the exchange of psychic energy, encouragement, meaning of life, and reasons to live, the long now is a boon.

That is not true about profits. Economically, the more the long tail expands, the more stuff there is to compete with our limited attention as an audience, the more difficult it is for a creator to sell profitably. Or, the longer the tail, the worse for sales.

The Web has significantly reduced the costs of producing and distributing content. Anyone with a computer can publish to a potential audience of hundreds of millions of people for as little as the cost of their Internet connection. This is great for content consumers but it has significantly increased the amount of competition among content creators while also reducing their chances of generating profits from their work since the Web/Internet has provided lots of options for getting quality content for free (both legally and illegally). 

All of this is a long way of saying that in the era of "Web 2.0" it was quite unwise for a VC funded startup to jump into the pool of content creators and thus become a victim of The Long Tail instead of becoming a content aggregator and thus benefiting from the Long Tail instead. Of course, even that may not have saved them since the market for podcast aggregators pretty much dried up once Apple entered the fray.

Now Playing: Lil Wayne - I'm Me

 

One of the problems you have to overcome when building a social software application is that such applications often depend on network effects to provide value to users. An instant messaging application isn't terribly useful unless your friends use the same application and using Twitter feels kind of empty if you don't follow anyone. On the flip side, once an application crosses a particular tipping point then network effects often push it to near monopoly status in certain social or regional networks. This has happened with eBay, Craigslist, MySpace, Facebook and a ton of other online services depend on network effects. Thus there is a lot of incentive for developers of social software applications to do their best to encourage and harness network effects in their user scenarios.

These observations have led to the notion of Viral Applications, applications which spread like viruses. The problem with a lot of the thinking behind "viral applications" and applications that borrow their techniques is that attempting to spread by any means necessary can be very harmful to the user experience. Here are two examples taken from this week's headlines

From Justine Ezaric, a post entitled The Loopt Debacle where she writes

Loopt is a location based social networking site that uses GPS to determine your exact location and share it with your friends.. and then spam your entire contact list via an SMS invite.

There’s a good chance that if you installed this application you’ve made the same mistake that most people made. While searching for friends who were on the service, apparently a text message was sent out to a large portion of my contact list, along with my phone number and my exact location (you know, since that’s the point of the application). Granted, you would think that if you have someone’s phone number, they’d have yours as well…

Hi, hey.. Over here!! People change their phone number for a reason!! With the ease of syncing contacts on the iPhone, it’s not always guaranteed that everyone in your contact list is a BFF (read: best friend forever). Also, there’s always people you just never want to text.. Like Steve Jobs, or an old boss, or maybe even an ex who would rather push you in front of a bus than get a text message from you?

From Marshal Kirkpatrick, a post entitled Gmail Tries To Be Less Creepy, Fails which states

Gmail, Google's powerful web based email service, announced some changes to its contact management features today. Contact management has for some time been a contentious matter among Google Account holders - the company does strange and mysterious things with your email contacts, including tying them in to some other applications without anyone's permission.

Today's new changes failed to alleviate those concerns, perhaps making the situation even less clear than it was before.

There Are Your Contacts and Then There Are Your Contacts

The post on the official Gmail blog today announced a new policy. There are now two types of contacts in your Gmail contacts list. There are your explicitly added My Contacts and there are your frequently emailed Suggested Contacts. The distinction between the two is unclear enough that I won't even try to summarize it. Read the following closely.

My Contacts contains the contacts you explicitly put in your address book (via manual entry, import or sync) as well as any address you've emailed a lot (we're using five or more times as the threshold for now).

Suggested Contacts is where Gmail puts its auto-created contacts. By default, Suggested Contacts you email frequently are automatically added to My Contacts, but for those of you who prefer tighter control of your address books, you can choose to disable usage-based addition of contacts to My Contacts (see the checkbox in the screenshot above). Once you do this, no matter how many times you email an auto-added email address it won't move to My Contacts.

…
When you open up Google Reader, the company's RSS reader, you'll find not just the feeds you've subscribed to but also the feeds of shared items from your "friends." Those friendships were defined somehow by Google, according to who you email in Gmail apparently. They can opt-out of having their shared items publicly visible at all, but short of doing that - you are seeing their shared items and someone, presumably, is seeing your shared items too. No one knows for sure.

Both Loopt and Gmail + Google Talk + Google Reader are examples of applications choosing approaches that encourage virality of the application or features of the application at the risk of putting users in socially awkward situations. As Justine mentions in the Loopt example, just because a person's phone number is in the contact list on your phone doesn't mean they would like to receive a text message from you at some random time of the day asking them to try out some social networking application. A phone isn't a social networking site. I have my doctor, my boss, his boss, our childcare provider, co-workers whose numbers I have in case of emergency and a bunch of other folks in my phone's contact list. These aren't the people I want to send spammy invites to try out some social networking application which probably doesn't even work on their phone. However I'm sure there has been some positive user growth from their "viral" techniques, but at what cost to their brand? Plaxo is still dealing with damage to their brand from their spammy era.

The Gmail behavior is even worse primarily because Google didn't fix the problem. Especially since people have been complaining about it for a while. No one can blame Google for wanting to jump start network effects for features like Shared Items in Google Reader or products like Google Talk, but it seems pretty ridiculous to decide to automatically add people I email to an IM application so they can see when I'm online and contact me anytime or to the list of people who are notified whenever I share something in Google Reader. It's just email, it does not imply an intimate social relationship. The worst thing about Google's practices is how it backfires, I'm less likely to use that combination of Google products so as not to cause inadvertent information leakage because some "viral algorithm" decided that because I sent a bunch of emails to my child care provider she needs to know whenever I share a link in Google Reader. 
 
If you decide to spread virally, you should be careful that you don't end up causing people to avoid your product like the diseases you are trying to emulate.

Now Playing: David Banner - Get Like Me (feat. Chris Brown, Yung Joc & Jim Jones)

 

About a week ago, the Facebook Data team quietly released the Cassandra Project on Google Code. The Cassandra project has been described as a cross between Google's BigTable and Amazon's Dynamo storage systems. An overview of the project is available in the SIGMOD presentation on Cassandra available at SlideShare. A summary of the salient aspects of the project follows.

The problem Cassandra is aimed at solving is one that plagues social networking sites or any other service that has lots of relationships between users and their data. In such services, data often needs to be denormalized to prevent having to do lots of joins when performing queries. However this means the system needs to deal with the increased write traffic due to denormalization. At this point if you're using a relational database, you realize you're pretty much breaking every major rule of relational database design. Google tackled this problem by coming up with BigTable. Facebook has followed their lead by developing Cassandra which they admit is inspired by BigTable. 

The Cassandra data model is fairly straightforward. The entire system is a giant table with lots of rows. Each row is identified by a unique key. Each row has a column family, which can be thought of as the schema for the row. A column family can contain thousands of columns which are a tuple of {name, value, timestamp} and/or super columns which are a tuple of {name, column+} where column+ means one or more columns. This is very similar to the data model behind Google's BigTable.

As I mentioned earlier, denormalized data means you have to be able to handle a lot more writes than you would if storing data in a normalized relational database. Cassandra has several optimizations to make writes cheaper. When a write operation occurs, it doesn't immediately cause a write to the disk. Instead the record is updated in memory and the write operation is added to the commit log. Periodically the list of pending writes is processed and write operations are flushed to disk. As part of the flushing process the set of pending writes is analyzed and redundant writes eliminated. Additionally, the writes are sorted so that the disk is written to sequentially thus significantly improving seek time on the hard drive and reducing the impact of random writes to the system. How important is improving seek time when accessing data on a hard drive? It can make the difference between taking hours versus days to flush a hundred gigabytes of writes to a disk. Disk is the new tape.

Cassandra is described as "always writable" which means that a write operation always returns success even if it fails internally to the system. This is similar to the model exposed by Amazon's Dynamo which has an eventual consistency model.  From what I've read, it isn't clear how writes operations that occur during an internal failure are reconciled and exposed to users of the system. I'm sure someone with more knowledge can chime in in the comments.

At first glance, this is a very nice addition to the world of Open Source software by the Facebook team. Kudos.

Found via James Hamilton.

PS: Is it me or is this the second significant instance of Facebook Open Sourcing a key infrastructure component "inspired" by Google internals?

Now Playing: Ray J - Gifts

 

Via Mark Pilgrim I stumbled on an article by Scott Loganbill entitled Google’s Open Source Protocol Buffers Offer Scalability, Speed which contains the following excerpt

The best way to explore Protocol Buffers is to compare it to its alternative. What do Protocol Buffers have that XML doesn’t? As the Google Protocol Buffer blog post mentions, XML isn’t scalable:

"As nice as XML is, it isn’t going to be efficient enough for [Google’s] scale. When all of your machines and network links are running at capacity, XML is an extremely expensive proposition. Not to mention, writing code to work with the DOM tree can sometimes become unwieldy."

We’ve never had to deal with XML in a scale where programming for it would become unwieldy, but we’ll take Google’s word for it.

Perhaps the biggest value-add of Protocol Buffers to the development community is as a method of dealing with scalability before it is necessary. The biggest developing drain of any start-up is success. How do you prepare for the onslaught of visitors companies such as Google or Twitter have experienced? Scaling for numbers takes critical development time, usually at a juncture where you should be introducing much-needed features to stay ahead of competition rather than paralyzing feature development to keep your servers running.

Over time, Google has tackled the problem of communication between platforms with Protocol Buffers and data storage with Big Table. Protocol Buffers is the first open release of the technology making Google tick, although you can utilize Big Table with App Engine.

It is unfortunate that it is now commonplace for people to throw around terms like "scaling" and "scalability" in technical discussions without actually explaining what they mean. Having a Web application that scales means that your application can handle becoming popular or being more popular than it is today in a cost effective manner. Depending on your class of Web application, there are different technologies that have been proven to help Web sites handle significantly higher traffic than they normally would. However there is no silver bullet.

The fact that Google uses MapReduce and BigTable to solve problems in a particular problem space does not mean those technologies work well in others. MapReduce isn't terribly useful if you are building an instant messaging service. Similarly, if you are building an email service you want an infrastructure based on message queuing not BigTable. A binary wire format like Protocol Buffers is a smart idea if your applications bottleneck is network bandwidth or CPU used when serializing/deserializing XML.  As part of building their search engine Google has to cache a significant chunk of the World Wide Web and then perform data intensive operations on that data. In Google's scenarios, the network bandwidth utilized when transferring the massive amounts of data they process can actually be the bottleneck. Hence inventing a technology like Protocol Buffers became a necessity. However, that isn't Twitter's problem so a technology like Protocol Buffers isn't going to "help them scale". Twitter's problems have been clearly spelled out by the development team and nowhere is network bandwidth called out as a culprit.

Almost every technology that has been loudly proclaimed as unscalable by some pundit on the Web is being used by a massively popular service in some context. Relational databases don't scale? Well, eBay seems to be doing OK. PHP doesn't scale? I believe it scales well enough for Facebook. Microsoft technologies aren't scalable? MySpace begs to differ. And so on…

If someone tells you "technology X doesn't scale" without qualifying that statement, it often means the person either doesn't know what he is talking about or is trying to sell you something. Technologies don't scale, services do. Thinking you can just sprinkle a technology on your service and make it scale is the kind of thinking that led Blaine Cook (former architect at Twitter) to publish a presentation on Scaling Twitter which claimed their scaling problems where solved with their adoption of memcached. That was in 2007. In 2008, let's just say the Fail Whale begs to differ. 

If a service doesn't scale it is more likely due to bad design than to technology choice. Remember that.

Now Playing: Zapp & Roger - Computer Love

 

I read two stories about companies adopting Open Source this week which give some interesting food for thought when juxtaposed.

The first is a blog post on C|Net from Matt Asay titled Ballmer: We'll look at open source, but we won't touch where he writes

Ballmer lacks the imagination to conceive of a world where Microsoft could open-source code and still make a lot of money (He's apparently not heard of "Google."):

No. 1, are our products likely to be open-sourced? No. We do provide our source code in special situations, but open source also implies free, free is inconsistent with paying for lunches at the partner conference. (Applause.)

But at least he's willing to work with those who do grok that the future of software business (meaning: money) is open source:

The second is an article on InfoWorld by Paul Krill entitled Sun lays off approximately 1,000 employees which contains the following excerpts

Following through on a restructuring plan announced in May, Sun on Thursday laid off approximately 1,000 employees in the United States and Canada. All told, the company plans to reduce its workforce by approximately 1,500 to 2,500 employees worldwide. Additional reductions will occur in other regions including EMEA (Europe, Middle East, Africa), Asia-Pacific, and Latin America. Reducing the number of employees by 2,500 would constitute a loss of about 7 percent of the company's employees.
...
He also addressed the question of whether Sun should abandon its new strategy of giving away its software. Sun will not stop giving it away, according to Schwartz, citing a priority in developer adoption.

When it comes to the financial benefits of Open Source, you need to look at two perspectives. The perspective of the software vendor (the producer) and the perspective of the software customer (the consumer). A key benefit of Open Source/Free Software to software consumers is that it tends to drive the price of the software to zero. On the other hand, although software producers like Sun Microsystems spend money to produce the software they cannot directly recoup that investment by charging for the software. Thus if you are a consumer of software, it is clear why Open Source is great for your bottom line. On the flip side, it isn't so clear if your primary business is producing software.

Matt Asay's usage of Google as an example of a company "making money" from Open Source is a prime example of this schism in perspectives. Google's primary business is selling advertising. Like every other media business, they gather an audience by using their products as bait and then sell that audience to advertisers. Every piece of software not directly related to the business of selling ads is tangential to Google's business. The only other software that is important to Google's business is the software that gives them a differentiated offering when it comes to gathering that audience. Both classes of software are proprietary to Google and always will be.

This is why you'll never find a Subversion source repository on http://code.google.com with the source code behind Google's AdSense or Adwords products or the current algorithms that power their search engine. Instead you will find Google supporting and releasing lots of Open Source software that is tangential its core business while keeping the software that actually makes them money proprietary. 

This means that in truth Google makes money from proprietary software. However since it doesn't distribute its proprietary software to end users, there isn't anyone complaining about this fact.

Unlike Google, Sun Microsystems doesn't really seem to know how they plan to make money. There is a lot of data out there that shows that the Sun Microsystems' model of scaling services is dying. Recently, Kai Fu Lee of Google argued that scaling out on commodity hardware is 33 times more efficient than using expensive hardware. This jibes with the sentiments of people who work on cloud services at Microsoft and Amazon that I've talked to when comparing the use of lots of "commodity" servers versus more expensive "big iron" server systems. This means Sun's hardware business is being squeezed because it is betting against industry experience. Giving away their software does not fix this problem, it makes it worse by cutting of a revenue stream as their core business is turning into a dinosaur before their eyes.

The bottom line is that giving something away that costs you money to produce only makes sense as part of a strategy that makes you even more money than selling what you gave away (e.g. free T-shirts with corporate logos). Google gets that. It seems Sun Microsystems does not. Neither does Matt Asay.

Now Playing: Inner Circle - Sweat (A La La La La Long)

 

When it comes to scaling Web applications, every experienced Web architect eventually realizes that Disk is the New Tape. Getting data from off of the hard drive disk is slow compared to getting it from memory or from over the network. So an obvious way to improve the performance of your system is to reduce the amount of disk I/O your systems have to do which leads to the adoption of in-memory caching. In addition, there is often more cacheable data on disk than there is space in memory since memory to disk ratios are often worse than 1:100 (Rackspace's default server config has 1GB of RAM and 250 GB of hard disk ). Which has led to the growing popularity of distributed, in-memory, object caching systems like memcached and Microsoft's soon to be released Velocity. 

memcached can be thought of as a distributed hash table and its programming model is fairly straightforward from the application developer's perspective. Specifically, There is a special hash table class used by your application which is in actuality a distributed hashtable whose contents are actually being stored on a cluster of machines instead of just in the memory of your local machine.

With that background I can now introduce Terracotta, a product that is billed as "Network Attached Memory" for Java applications. Like distributed hash tables such as memcached, Terracotta springs from the observation that accessing data from a cluster of in-memory cache servers is often more optimal than getting it directly from your database or file store.

Where Terracotta differs from memcached and other distributed hash tables is that it is completely transparent to the application developer. Whereas memcached and systems like it require developers to instantiate some sort of "cache" class and then use that as the hash table of objects that should be stored, Terracotta attempts to be transparent to the application developer by hooking directly into the memory allocation operations of the JVM.

The following is an excerpt from the Terracotta documentation on How Terracotta Works

Terracotta uses ASM to manipulate application classes as those classes load into the JVM. Developers can pick Sun Hotspot or IBM's runtime, and any of several supported application servers
…
The Terracotta configuration file dictates which classes become clustered and which do not. Terracotta then examines classes for fields it needs to cluster, and threading semantics that need to be shared. For example, if to share customer objects throughout an application cluster, the developer need only tell Terracotta to cluster customers and to synchronize customers cluster-wide.

Terracotta looks for bytecode instructions like the following (not an exhaustive list):

• GETFIELD
• PUTFIELD
• AASTORE
• AALOAD
• MONITORENTRY
• MONITOREXIT

On each of those, Terracotta does the work of Network Attached Memory. Specifically:

BYTECODE	Injected Behavior
GETFIELD	Read from the Network for certain objects. Terracotta also has a heap-level cache that contains pure Java objects. So GETFIELD reads from RAM if-present and faults in from NAM if a cache miss occurs.
PUTFIELD	Write to the Network for certain objects. When writing field data through the assignment operator "=" or through similar mechanisms, Terracotta writes the changed bytes to NAM as well as allowing those to flow to the JVM's heap.
AASTORE	Same as PUTFIELD but for arrays
AALOAD	Sames as GETFIELD but for arrays
MONITORENTRY	Get a lock inside the JVM on the specified object AND get a lock in NAM in case a thread on another JVM is trying to edit this object at the same time
MONITOREXIT	Flush changes to the JVM's heap cache back to NAM in case another JVM is using the same objects as this JVM

The instrumented-classes section of the Terracotta config file is where application developers specify which objects types should be stored in the distributed cache and it is even possible to say that all memory allocations in your application should go through the distributed cache.

In general, the approach taken by Terracotta seems more complicated, more intrusive and more error prone than using a distributed hash table like Velocity or memcached. I always worry about systems that attempt to hide or abstract away the fact that network operations are occurring. This often leads to developers writing badly performing or unsafe code because it wasn't obvious that network operations are involved (e.g. a simple lock statement in your Terracotta-powered application may actually be acquiring distributed locks without it being explicit in the code that this is occuring).

Now Playing: Dream - I Luv Your Girl (Remix) (feat. Young Jeezy)

 

In the past year both Google and Facebook have released the remote procedure call (RPC) technologies that are used for communication between servers within their data centers as Open Source projects. 

Facebook Thrift allows you to define data types and service interfaces in a simple definition file. Taking that file as input, the compiler generates code to be used to easily build RPC clients and servers that communicate seamlessly across programming languages. It supports the following programming languages; C++, Java, Python, PHP and Ruby.

Google Protocol Buffers allows you to define data types and service interfaces in a simple definition file. Taking that file as input, the compiler generates code to be used to easily build RPC clients and servers that communicate seamlessly across programming languages. It supports the following programming languages; C++, Java and Python.

That’s interesting. Didn’t Steve Vinoski recently claim that RPC and it's descendants are "fundamentally flawed"? If so, why are Google and Facebook not only using RPC but proud enough of their usage of yet another distributed object RPC technology based on binary protocols that they are Open Sourcing them? Didn’t they get the memo that everyone is now on the REST + JSON/XML bandwagon (preferrably AtomPub)?

In truth, Google is on the REST + XML band wagon. Google has the Google Data APIs (GData) which is a consistent set of RESTful APIs for accessing data from Google's services based on the Atom Publishing Protocol aka RFC 5023. And even Facebook has a set of plain old XML over HTTP APIs (POX/HTTP) which they incorrectly refer to as the Facebook REST API.

So what is the story here?

It is all about coupling and how much control you have over the distributed end points. On the Web where you have little to no control over who talks to your servers or what technology they use, you want to utilize flexible technologies that make no assumptions about either end of the communication. This is where RESTful XML-based Web services shine. However when you have tight control over the service end points (e.g. if they are all your servers running in your data center) then you can use more optimized communications technologies that add a layer of tight coupling to your system. An example of the kind of tight coupling you have to live with is that  Facebook Thrift requires specific versions of g++ and Java if you plan to talk to it using code written in either language and you can’t talk to it from a service written in C#.

In general, the Web is about openness and loose coupling. Binary protocols that require specific programming languages and runtimes are the exact opposite of this. However inside your Web service where you control both ends of the pipe, you can optimize the interaction between your services and simplify development by going with a binary RPC based technology. More than likely different parts of your system are already doing this anyway (e.g. memcached uses a binary protocol to talk between cache instances, SQL Server uses TDS as the communications protocol between the database and it's clients, etc).

Always remember to use the right tool for the job. One size doesn’t fit all when it comes to technology decisions.

FURTHER READING

• Exposing a WCF Service With Multiple Bindings and Endpoints – Keith Elder describes how Windows Communication Foundation (WCF) supports multiple bindings that enable developers to expose their services in a variety of ways.  A developer can create a service once and then expose it to support net.tcp:// or http:// and various versions of http:// (Soap1.1, Soap1.2, WS*, JSON, etc).  This can be useful if a service crosses boundaries between the intranet and the Internet.

Now Playing: Pink - Family Portrait

 

A year ago Loren Feldman produced a controversial video called "TechNigga" which seems to still be causing him problems today. Matthew Ingram captures the latest fallout from that controversy in his post Protests over Verizon deal with 1938media where he writes

Several civil-rights groups and media watchdogs are protesting a decision by telecom giant Verizon to add 1938media’s video clips to its mobile Vcast service, saying Loren’s "TechNigga" clip is demeaning to black people. Project Islamic Hope, for example, has issued a statement demanding that Verizon drop its distribution arrangement with 1938media, which was just announced about a week ago, and other groups including the National Action Network and LA Humanity Foundation are also apparently calling for people to email Verizon and protest.

The video that has Islamic Hope and other groups so upset is one called "TechNigga," which Loren put together last August. After wondering aloud why there are no black tech bloggers, Loren reappears with a skullcap and some gawdy jewelry, and claims to be the host of a show called TechNigga. He then swigs from a bottle of booze, does a lot of tongue-kissing and face-licking with his girlfriend Michelle Oshen, and then introduces a new Web app called "Ho-Trackr," which is a mashup with Google Maps that allows prospective johns to locate prostitutes. In a statement, Islamic Hope says that the video "sends a horrible message that Verizon seeks to partner with racists."

I remember encountering the video last year and thinking it was incredibly unfunny. It wasn’t a clever juxtaposition of hip hop culture and tech geekery. It wasn’t satire since that involves lampooning someone or something you disapprove off in a humorous way (see The Colbert Report).  Of course, I thought the responses to the video were even dumber; like Robert Scoble responding to the video with the comment “Dare Obasanjo is black”.

Since posting the video Loren Feldman has lost a bunch of video distribution deals with the current Verizon deal being the latest. I’ve been amused to read all of the comments on TechCrunch about how this violates Loren’s freedom of speech.

People often confuse the fact that it is not a crime to speak your mind in America with the belief that you should be able to speak your mind without consequence. The two things are not the same. If I call you an idiot, I may not go to jail but I shouldn’t expect you to be nice to me afterwards. The things you say can come back and bite you on butt is something everyone should have learned growing up. So it is always surprising for me to see people petulantly complain that “this violates my freedom of speech” when they have to deal with the consequences of their actions.

BONUS VIDEO: A juxtaposition of hip hop culture and Web geekery by a black tech blogger.

Now Playing: NWA – N*ggaz 4 Life