tech vs non-tech startups

Technological start-up means technological value given to customer, google an example, search technology. An opposite facebook is not so tech company, but social and mass oriented, twitter easy to create a clone but who need this?
The real challenge is not just to use current technologies but extend or create new ones.

cloud computing rocks, especially for DIRTy apps

Cloud technology bum! Especially cloud databases.
Projects to maintain infrastructure:

openstack eucalyptus opennebula etc. and hands :)

Distributed file systems:

hdfs glusterfs ceph lustre

Cloud databases:
mongodb riak couchdb swift.openstack redis  hypertable hbase cassandra etc.

Frameworks:

hadoop storm mesos spark mpi cascading

Computing patterns:

mapreduce distributed-rpc

Cloud tools:
yahoo pipes
google publicdata
google fusiontables

Real time re-configurable network topology is critically important to create data intensive real time #DIRTy apps. Traditional computation schema is based on application graph, where vertices are mapped to computing nodes and  ages represent network connections between nodes. So what's wrong with it?

It has potential bottlenecks as any data-flow computing if too many data go thorough one path. Here is well known graph mapping task, two graphs one is application graph, second is network & computing resources one. Initial task is to create effective mapping of application graph to cloud network topology. Real thing to create fault-tolerant continuous cloud computing  of DIRTy applications.

#startups are shaping the world

Startups are so popular topic for many years, but this days especially IT-startups too much buzzing. I thought winter holidays slow down this very active  process, I was wrong. I see new wave is coming :)
The problem is that as usual every #startup must or should has an investor. This is important, you see the following? startup follows investor. Elevator pitch, business plan, presentation, prototype, MVP... all know this stuff, isn't it? This really wrong way.
Two things:

• newer follow others so hard, let others follow you;
• investors use startups to shape the world. 

Yes startups shape the world, don't agree? just google, not to google, just the google :)

But startups are selected and managed by investors, so investors indirectly shape the world!

What is then alternative? Bootstrapinpg ? Not stand-alone bootstrapping.

As an analyst I could say, yes this movement should go! Do not wait till all money cats rebuild the world, just create alternate freedom of choice.

New kind of computing on Open Data

#OpenData  just yet another buzzword like web 2.0, semantic web, etc. But behind the scenes we could see much more potential and opportunities. What's difference between  #OpenData  and just familiar web blogs for instance? First of all it's representation & organization, #OpenData  is representation agnostic computable thing from one side & perfectly human readable & "viewable" in many forms in other side. Semantic web? So semantic web is dead, #OpenData  comes instead :). Twitter is one of the first shy steps to that direction of #OpenData concept, so is it hell or heaven? Both, depends on us. We should clearly understand & use it for good.
I'm thinking to support two options in registration form:
- human with captcha test (programs not yet able to pass);
- program with heavy calculation test (humans won't be able to pass);
Crazy? No, may be just some fun. Real reason is that #OpenData really nice opportunity to create new kind of computing engines on top of it.
To be continued...

cloud tech

Cloud technologies very hot topic today, most buzzing is cloud non-relational databases. Pretty interesting ones are following

• mongodb (10gen) 
• dynamodb (amazon)
• sensei db (linkedin)
• riak (basho)

Database as a Service, some opensource databases are used by 3d parties for instance mongolab or proprietary like dynamodb is available recent time. Monsters like google offers cloud storage & databases also including cloudsql.

What business model, just like any other XaaS services different hosting plans with periodical payments.

multi language open data

It's exciting time to participate in opendata research & developmet. Yes It is revolution, no turn back. The humans at last come close as never before to open data. So wikipedia, twitter, socrata, etc. & a great number of data dot gov resources!
Open data about everything, just little example with  wikipedia style - many languages!
{
 en : {
  "message" : "opendata is cool",
  "created_at" : "Fri Jan 13 2012 01:51:41 GMT+0400 (MSK)",
  "tags" : ["open","data","cool"],
 }
 ru : {
  "сообщение" : "открытые данные это круто",
  "дата_создания" : "Пт Янв 13 2012 01:51:41 GMT+0400 (MSK)",
  "тэги" : ["открытые","данные","круто"],
 }

 ...

}

opendata+opencode=opendataobjects

#opendata much spoken concept, but just data is not enough. Yes + the code, it's critically important!
The REST architectural style widely used today has no data format,  as usual it's JSON or XML but no matter what it is. JSON is the best choose, easy for humans to use, easy for programms to manipulate.
Ok, just data is not enough, data + code. What code? Code what for?
Two things:
- open source code is not ready to use immediately, you have to build it as minimum, resolve dependencies, create config files, etc. But imagine what if any one can run open source software as a service (OSSaaS), no make files, no linux stuff, just put query in a browser address line, no programming skills required!
- just data is not enough, who put this data to open data cloud? Men in black, doubt it. Humans manually upload exel files there, omg. An idea to make opendataobjects live, just add code!
opendataobjects=opendata+opencode!
How it might looks like, no bicycle inventing, just in JSON:
{
  field1 : "value1",
  field2 : "value2",
  ...
  field_method1: "script code",
  field_method1_lang: "javascript",


  ...
  field_method2: "script code",
  field_method2_lang: "python",

  ...

}

Opendataobjects is a concept to represent opendata with open methods to play with this opendata. Yes we know all this OOP stuff, lets take what we need here!
You put in wikipedia just static text, let's imagine you put live opendataobjects here!

Who's responsible for execution of opendataobjects methods, both open data cloud providers and 3d party services, this code can do any http calls, send messages to other objects, etc.

Lets make cloud more open! 

..to be probably continued :)