Upload
gaurav-mishra
View
213
Download
0
Embed Size (px)
DESCRIPTION
Cloud
Citation preview
By Kari JohnsonSpecial Correspondent
Oil and gas operations present severalchallenges to mobile computing hardwareand software, ranging from the types andvolumes of data, to the simple fact thatmobile devices have to be able to tolerateextreme conditions. A shiny off-the-shelfcommercial tablet probably would havea very limited life expectancy on a busydrilling rig or production site, and themobile applications written for the typicalconsumer likely would be grossly inade-quate for oil and gas, given the complexnature of the workflows.
However, the basic hardware and soft-ware components to make mobile andcloud computing a reality in explorationand production are rapidly coming tomarket, with new devices and applicationoptions available seemingly by the day.A case in point is a variety of purpose-engineered ruggedized tablets that aredesigned to withstand the elements presentin even the worlds harshest operatingextremes, including Arctic climates andhazardous environments.
Cloud Solutions
The hardware is only part of the picture,of course. Longer term, cloud computingsolutions are expected to play a centralrole in allowing mobile computing toreach its full potential in upstream oiland gas. One of the most appealing aspectsof cloud computing is also what makesthe concept so difficult for potential oiland gas users to wrap their heads around,according to Barbara Murphy, chief mar-keting officer at Panasas.
It seems that the cloud is whateveryou want it to be, she states. Clouds
are engineered to provide accessibilityanywhere on demand, and are both highlyscalable and highly virtualized. Thesekey attributes are consistent betweenpublic and private cloud definitions. Infact, the difference between public andprivate is almost irrelevant to users.
From the Panasas perspective, thevalue driver for the oil and gas industryis in cloud computings potential to bringtogether massive compute and storagepower for far better performance and ef-ficiency. In this industry, we have to de-sign for peak loads. Many smaller inde-pendent compute clusters, each designedfor peak load, can never be as economicalas one centralized resource, Murphy ex-plains.
In computing circles, the oil and gasindustry is synonymous with massivedata processing, storage and visualizationrequirements, according to Murphy. While
storage is certainly a key part of the costequation, she says the biggest price tagis associated with the cost of the CPUs.
In a typical high-performance clusterenvironment with 10,000-30,000 cores,storage is about 25 percent of the totalsystem cost, while compute cores can becloser to 50 percent, notes Murphy. Oneof the major drivers behind private cloudsis the desire to make the most efficientuse of the cores, which represent the ma-jority of the capital investment in high-performance clusters. If you can improvethe utilization of compute resources from40 to 80 percent across a company whilestill delivering superior performance, thesavings are tremendous.
Private clouds also enable consolida-tion. Murphy gives the example of Leices-ter University, which had multiple indi-vidual departments managing smallerhigh-performance computing systems for
Technologies Providing New Solutions
Cloud computing offers the potential to bring together massive compute and storagepower for improved performance and efficiency in the oil and gas industry. Private cloudsimplemented behind a companys firewall can provide tremendous savings by improvingthe utilization of computing and storage resources across the organization while stilldelivering superior performance.
The Better Business Publication Serving the Exploration / Drilling / Production Industry
NOVEMBER 2011
Reproduced in part and adapted for Panasas with permission from The American Oil & Gas Reporter
www.aogr.com
www.aogr.comscientific applications. By aggregatingthem into a single managed resource, itallowed the university to gain enormousefficiencies and significant cost savings,she says.
Two other trends impacting privatecloud computing are segregating high-and low-value computing jobs, in whichthe highest-value hardware and softwareare deployed to the highest-value jobs,and parallelization of applications tocomplete computational-intensive taskssignificantly faster. Both of these trendsare driven by the desire to solve complexbig data computing problems faster andmore cost efficiently, Murphy details.The next generation of computing willbe all about parallelism, in particular.
As far as the public cloud, Murphy
says users primarily are looking for costefficiencies for general purpose computingand data storage capacity. At the sametime, they are very clear about not wantingto process massive data sets or sensitivedata in the public cloud, she observes.Not only are they concerned about se-curity, but data sets that are tens to hun-dreds of terabytes in size simply cannotmove across a wide-area network in acost-effective manner. More importantly,applications with a high data-to-computeratio remain a challenge for public clouds,since they cannot deliver the performancerequired.
Storage systems serving a high-per-formance private cloud must have severalkey characteristics, starting with a dedi-cated high-bandwidth connection between
the compute resources and the storagesystem, Murphy advices. Storage alsomust be continuously available and havethe ability to scale out seamlessly withoutever going offline.
For budgeting assistance and planning,tools are available to manage user quotasso that costs can be allocated based onusage rather than simple head count. Thisquota system should be a soft allocation,so that users do not hit a limit that stopswork, she suggests. Good managementtools are also essential. It takes a lot ofthought on how the user experience willactually manifest itself. The storage man-agement systems must be both a billingdepartment and a police departmentagainst potential security threats. r
SpecialReport: Oil & Gas Computing