Skip to main content

Hadoop engineering support services primer

By Sachin Ghai
As the focus has shifted to production deployment of Hadoop based applications, there is an increasing need to have good practices around the production support process. While this may be a BAU transition process for large organizations, the startups and new product orgs. may want to catch up on standard and best implementations. Let’s put across a small primer on what you need while setting up a Hadoop production support system.



People:


            In a typical software product org and/or startup, the first focus and strength is usually skilled people. However, there is a need to segregate your high skilled force among the product/project development and support activities. At best, you need the following skills as part of support org. structure:

-          Hadoop stack Administrator

-          Support professional with MapReduce and essential application knowledge

-          Data scientist (definition of skills may vary specific to your app)

-          Account Manager (multi-variant customer facing role)



Process:


            While process terms may sound slightly offbeat or cliché to ‘hackers’, a service management setup demands it as a basic commandment. To begin with, identify type of issue reported as:

-          Incident, Problem, Service Request

-          Product enhancement

Further, establish how your team can get access to stacks:

-          Logs

-          Remote login rights

At the onset, while taking an issue report, collect basic hygiene information from the customer, including

-          stack details with product versions

-          trigger for issue, if known

-          any new deployment or upgrade info in the cluster

-          data sample, if possible

Further, devise a process on how to:

-          Escalate ticket internally

-          Advise customer on escalation process for:

o       Hierarchical escalation within your organization

o       Cross team escalation for integrated stacks

o       ‘War room’ escalations for major revenue impact incidents



Technology:


            Technology is not limited to just your product and the associated hardware. For a production support system, some must-have include:

-          Voice Call response and recording

-          Ticket management portal for accepting, assigning and transferring tickets

It may be nice to have

-          Self service portal for customers

-          FAQ or answer forum (like MapR) to help customer figures out a few issues themselves



Metrics:


            No support team can be spared on this. You just ought to have a sound metrics system for your team on:

-          Turn Around Time

-          Effort

-          Cost

-          Post production defects

-          Productivity per person



Billing:


            The bread and butter step, this has a few catches when it comes to Hadoop stacks. Your account manager (aka VP) needs to have a clear cut contract on support cost per node or similarly defined unit along with billing milestones. So, you need to keep watching each month for:

-          Any contract deviations (e.g. from inclusions in fixed AMC)

-          Number of supported nodes that your contract stated

-          Any ticket which needs to be billed as Enhancement and not as problem.





While many of these may inspire from a usual service management desk, there are key peculiarities with node based billing combined along with evolutionary nature of MapReduce product ecosystem. Due to the integrated nature of Hadoop technology stack which may have unsupported open source along with evolving NoSQL database combined with enterprise legacy suites, each organization needs to evolve a resilient production support system. It is best advised to review the key constituents holistically and devise a production support organization system.



For any help in setting up a production support setup further, you may drop a query to scale [at] hadoopsphere.com .





 ______________________

 top image source: freedigitalphotos.net
Labels:

Comments

Post a Comment

Popular posts from this blog

In-memory data model with Apache Gora

By Sachin Ghai
Open source in-memory data model and persistence for big data framework Apache Gora™ version 0.3, was released in May 2013. The 0.3 release offers significant improvements and changes to a number of modules including a number of bug fixes. However, what may be of significant interest to the DynamoDB community will be the addition of a gora-dynamodb datastore for mapping and persisting objects to Amazon's DynamoDB. Additionally the release includes various improvements to the gora-core and gora-cassandra modules as well as a new Web Services API implementation which enables users to extend Gora to any cloud storage platform of their choice. This 2-part post provides commentary on all of the above and a whole lot more, expanding to cover where Gora fits in within the NoSQL and Big Data space, the development challenges and features which have been baked into Gora 0.3 and finally what we have on the road map for the 0.4 development drive.
Introducing Apache Gora Although there are var…
4 comments
Read more

Amazon DynamoDB datastore for Gora

By Sachin Ghai
What was initially suggested during causal conversation at ApacheCon2011 in November 2011 as a “neat idea”, would soon become prime ground for Gora's first taste of participation within Google's Summer of Code program. Initially, the project, titled Amazon DynamoDB datastore for Gora, merely aimed to extend the Gora framework to Amazon DynamoDB. However, it seem became obvious that the issue would include much more than that simple vision.

The Gora 0.3 Toolbox We briefly digress to discuss some other noticeable additions to Gora in 0.3, namely: Modification of the Query interface: The Query interface was amended from Query<K, T> to Query<K, T extends Persistent> to be more precise and explicit for developers. Consequently all implementors and users of the Query interface can only pass object's of Persistent type. Logging improvements for data store mappings: A key aspect of using Gora well is the establishment and accurate definitio…
1 comment
Read more

Data deduplication tactics with HDFS and MapReduce

By Desk Editor
As the amount of data continues to grow exponentially, there has been increased focus on stored data reduction methods. Data compression, single instance store and data deduplication are among the common techniques employed for stored data reduction.
Deduplication often refers to elimination of redundant subfiles (also known as chunks, blocks, or extents). Unlike compression, data is not changed and eliminates storage capacity for identical data. Data deduplication offers significant advantage in terms of reduction in storage, network bandwidth and promises increased scalability.
From a simplistic use case perspective, we can see application in removing duplicates in Call Detail Record (CDR) for a Telecom carrier. Similarly, we may apply the technique to optimize on network traffic carrying the same data packets.
Some of the common methods for data deduplication in storage architecture include hashing, binary comparison and delta differencing. In this post, we focus on how MapReduce and…
Post a Comment
Read more