Friday, 10 June 2011

Conference On Cloud Computing

Recently I attended 2nd IndicThreads Conference On Cloud Computing.

About IndicThreads :
IndicThreads is one of India’s leading tech+media brands and has been successfully running the online magazine IndicThreads.com & hosting some of India’s finest independent technology conferences.

Speakers talked about various cloud platforms, cloud services in various forms, security aspects of cloud, cloud testing, monitoring your application in the cloud etc.

Here are a few highlights:

Scalable Architecture on Amazon AWS Cloud - Amazon Elastic Loadbalancing distributes incoming traffic to your app across several EC2 instances and reroutes traffic accordingly. Auto-scaling is enabled by Amazon CloudWatch. Amazon SimpleDB is schema-less with distributed key-value store and is scalable, has auto-indexing and has API based search. AWS Relational Database Service (RDS) lets you scale your DB layer with minimum administration. MySQLand Oracle are supported and you can import existing databases without any changes to app.

OpenStack Ecosystem and Xen Cloud Platform - Main idea behind OpenStack was to open source cloud computing platform that is simple to
implement and massively scalable. Available as OpenStack Compute(Nova), Storage(Swift) and Image Service(Glance). Openstack is built on a shared-nothing, messaging-based architecture using AMPQ based queues.
Xen Cloud Platform(XCP) is open source platform to build clouds, a cloud optimised version of Xen and has virtualization platform that consists of Xen hypervisor, network and storage support.
OpenStack supports XCP through XenAPI. OpenStack Compute interacts with the XCP hypervisor (XEN).

Processing massive amount of data with Map Reduce using Apache Hadoop - Map is applying a function to a list of elements and returning the modified list. Reduce/Fold reduces a list of values to a single value/element. With Hadoop MapReduce, you need to create your Mapper, Combiner(when the function to be applied is commutative and associative) to minimize the number of key value pairs that will be shuffled across the network between mappers and reducers, Partitioner that controls the partitioning of the keys of the intermediate map-outputs and Reducer

Easy ORM-ness with Objectify-Appengine- Entities, Operations(get, put, delete and query), Keys and Transactions form the basics of any datastore. Objectify supports these native features and provides type safe key and query classes. Objectify is light weight, maintains simplicity and transparency of low level API and does the work of converting to and from POJOs to Entity Objects.Certain practices will make Objectifying very easy: use of DAO to register entities, scanning only on demand, using batch gets rather than queries, lower usage of indexing and avoiding @Parent.

Cloud security - In public clouds you need to have strong identity management, secure keys, access to compliance and logging, wipeout policies. In private ones scope should be left for moving to a hybrid model and imposing SSO is preferred practice. Using hybrid ones segregate your data between clouds, use SAML, hHave your Federation set up correctly, enable SSO and ensure that enterprise credentials do not move into the cloud. Safe Browsing API and VM Safe API are the APIs useful in cloud security.

Cloud Lock-in and Interoperability - Locking in with a cloud would mean higher costs being paid and incompatible technology, for the customer. Portability will mean choice to move out of cloud or another cloud without losing any data or even the log-files. There are standards like Distributed Management Task Force(DTMF), Storage Networking Industry Association(SNIA) and Cloud Security Alliance(CSA).

Cloud development with CloudFoundry - CoudFoundry overcomes the issues that exists in present PaaS i.e. Google App Engine like: lack of cloud portability, tightly coupled with a certain environment and deployment on public clouds only. CloudFoundry is open source and with it you can to deploy on public or private cloud.

Monitoring Cloud Applications- There are various challenges at various levels involved such as at user, system and network level. You should be able to scale your app on cloud dynamically with no impact on monitored infrastructure. Monitoring tools should be portable and light weight.

AWS Elastic Beanstalk - Gives a handy way to deploy apps especially the ones using Tomcat stack. You have complete control over your AWS resources. Hence you can choose any database, access your Ec2 instances, create custom AMIs as well as move app out from Beanstalk with great ease. Apart from the command line tools and APIs, you can access Beanstalk using AWS Toolkit for Eclipse and AWS Management Console.

Apigee - API Management Platform : Apigee provides free API tools for developers and is the leading API management platform for enterprises.

loggly - Log Management in the Cloud : Loggly is a cloud based logging service. With Loggly, you can collect logs from your servers and then quickly search them with an intuitive user interface. We make looking at logs fast, fun and easy!

No comments:

Post a Comment