We were reviewing DynamoDB. We will compare it with Cosmos DB but first let us review Cosmos DB. Azure Cosmos DB was written to be the industry's first globally-distributed, multi-model database services. It is built to collect data from a plethora of IoT devices as well as to be handle the processing of machine-learning. It is a globally distributed data service that can elastically scale throughput and storage with data consistency, low latency and high availability. There are five different consistency levels which include strong, bounded-staleness, session, consistent-prefix and eventual consistency levels. No schema or index is required because the database automatically indexes all data and the engine does not know or care about any schemas. Key-value, documents and graph data models are easily supported.  
Perhaps one of the undeniably desirable feature of Cosmos DB is for developers to choose the data model and the API. Developers are no longer bound to one model.  They can easily switch from key-value to graph model as they write one application after another. This makes this DB  second nature to the developers and earns mindshare which is just as important as acquiring data because the two are related. Similarly they can also manipulate the data with the same set of APIs that they are already familiar with. This database does not impose any new APIs We can use the same dialects such as that of Document DB, Mongo DB and APIs of Gremlin database. These are in addition to the native Azure Table Storage API. Not only this but other data access APIs can be added to the offerings going forward.  Azure Cosmos is the first and only globally distributed database service in the industry to financially backed comprehensive SLAs  as per their documentation. This gives us the assurance that the APIs will not only be easier to work with but also highly available with low latency, high consistency and throughput. In other words, customers who were using NoSQL database can now for go their own maintenance and adopt a managed Database as a Service layer with little or no disruption to their existing applications. In our discussion of Dataset architecture here, we saw important concepts around new trends but nothing seems to bring it all in together as this managed service platform from Azure Cosmos. 

A text summarization attempt:  http://scratchpad.io/scintillating-spot-7410
API
UI

#codingexercise
Given a rectangular matrix with rowwise and columnwise sorted distinct integers, find a given integer
Tuple<int,int> GetPosition(int[,] A, int Rows, int Cols, int X)
{
var ret = new Tuple<int, int>(){-1,-1};
// x, y coordinates
for (int i =0; i < Rows; i++)
{
int index = binary_search(A, cols, i, X);
if ( index != -1)
{
ret.first = i;
ret.second = index;
break;
}
}
return ret;
}
We can also do binary search along columns as well as rows.