Infrastructure as a Service,” held October 22, 2010. The speaker was Kyle
Johnson, Associate Provost for Information and Learning Resources at the
SUNY Institute of Technology.
Many institutions have begun movingapplication services to the cloud, but whatother kinds of services could move there?
What might our institutions look like if desktops,storage, servers, and even the networkwere delivered via the cloud? This webinar
explores the possibilities for infrastructureas a service.
This webinar uses the following definitionof cloud computing:
“…the term cloud computing refers tothe delivery of scalable IT resourcesover the Internet, as opposed to hostingand operating those resources locally,such as on a college or university network.Those resources can includeapplications and services, as well as theinfrastructure on which they operate.”(EDUCAUSE, 7 Things You Should Knowabout Cloud Computing)
Cloud computing allows an institution topurchase a service level rather than softwareand hardware, theoretically making it easy to
scale services as needed. (Contractual obligationsare a negotiation and policy issue.)Most discussions of cloud computing focus
on software as a service, but the cloud alsooffers options for infrastructure as a service(IaaS), defined as:
“‘Infrastructure as a Service’ is a provisionmodel in which an organization outsources the equipment used to supportoperations, including storage,hardware, servers and networking components.The service provider owns theequipment and is responsible for housing,running and maintaining it.” (FromSearchCloudComputing.com)The two definitions are similar, but the focusin IaaS is more on the nuts and bolts of theoverall campus IT function through purchasinga service level, not hardware and software.
Steps in Migrating to IaaS
As a first step, an institution might bringin outside resources to assist in managingexisting infrastructure. This is essentially traditionaloutsourcing of support, with theinstitution still owning all the hardware andmaking all the major decisions about how tosupport the infrastructure.Next, the institution might want to partnerwith a company that can provide the expertiseto design, build, and manage infrastructure.At this point, the institution still owns all the hardware and probably the maintenancecontracts, but has moved select staffingresources elsewhere. This stage might reapmore benefits as the service provider suppliesmore hours of service. (At institutionswhere budget processes or policies make itdifficult to exchange capital dollars for operatingdollars, this might be the endpoint inadopting IaaS.)
In the last phase, the institution stops purchasingeven capital hardware, instead negotiatingwith a service provider for a given infrastructure.In this case, the institution is bounded primarilyby funding concerns and does not have the supplychain management concerns that commonlyaccompany investments in capital hardware
Note that at no time in adopting IaaS does aninstitution surrender the ability or responsibilityto set strategic directions for its infrastructure.The implementational and operational detailsmight change, but the infrastructure still needsto meet institutional priorities.
Considerations
There are a number of factors to weigh inconsidering IaaS.
First, there are two meaningful prerequisites.
Institutions need a well-designed andwell-implemented identity managementplan and system. When systemsare all on site, staff can manuallymanage identity and access controlsbetween systems. But as infrastructureis shifted to the cloud, an institutioncan’t rely on those solutions.
Institutions also need a robust, redundantnetwork. If an institution is bandwidth-constrainedor has problemswith performance, it will face challengesgetting the service level it needsfrom the cloud.
Second, there are financial considerations.Moving to the cloud model tendsto shift expenses from capital to operational.
If an institution doesn’t have theflexibility to shift dollars, IaaS will beharder to implement.„ Third, IaaS brings the same legal considerationsas software as a service. It isimportant to understand where data andinfrastructure will reside—that is, in whatstate or country—and whether an institutionwill be comfortable with where dataresides. Ensure also that IaaS contracts
are constructed to meet federal, state,and local regulations.
A fourth consideration is staffing. InIaaS, staff will need more expertise inprocurement, contract negotiation, andmanaging vendor relationships. Staffwill also devote more time to interactingwith the institutional community in needsgathering and analysis. Similarly, staff willspend more time on the integration ofservices. Migrating to IaaS requires differentstaff skills, so IT administrators needto be careful about how they train andtransition staff.
A final consideration is cultural. In shiftinginfrastructure to the cloud, it’s importantto involve members across the nstitutionalcommunity, both to fully understandtheir needs and to keep them apprised ofchanges that IaaS might bring. IaaS also
requires a shift in culture within the IT division,perhaps including a shift away fromtailoring its own solutions when solutions
already exist in the cloud. IT also needs torealize that it will lose some operationalcontrol, although that loss will be somewhat
offset by the ability gained via IaaSto better meet community needs.
Examples of IaaS
Next, consider examples of infrastructureas a service
Compute Cycles
Research into computing support can bechallenging. Often there is uneven demandfor the service, a large capital outlay required
to get it started, and specialized staff neededto maintain it. This kind of service, whendone in-house, doesn’t scale down very well.Moving infrastructure to the cloud allows aninstitution to provision and pay for computecycles on an as-needed basis, without theneed for staff on site to maintain them.A good example of this is Amazon ElasticCompute Cloud (EC2), which provides an application programming interface (API) sothat institutions and organizations can createa simple window that allows researchers toaccess compute cycles as needed, with severaloptions for paying for the service.
Servers
Many challenges are inherent in servermanagement. Many campuses host multipleoperating systems that require specialized
in-house expertise. Space, HVAC, and powerneeds can be substantial, even for a smallschool. Providing failover at a remote locationcan be expensive if an institution has toduplicate its entire service structure, or it canrequire difficult decisions about which serviceswon’t be duplicated.
Moving this infrastructure to the cloudlets an institution select a core operatingsystem based on need rather than the availabilityof in-house knowledge. The providerwill be responsible for failover and oftenalready has the necessary facilities and plansin place. With this solution, an institution canalso avoid building another new data centerinfrastructure.
To phase in this kind of service, an institutionmight start with remote support forsome of the systems for which it lacks on-site
expertise. Next, the institution might look intoheavier virtualization and engage a partnerto design, build, and manage the infrastructure.After that, a logical final step would beto migrate all virtual infrastructure off site
Amazon EC2 is an example of this service,and with the addition of the Virtual PrivateCloud extension, provides this infrastructure
on an isolated network with a virtualprivate network connection back to campus.Rackspace and GoGrid are designed for
permanent server installations at an off-sitelocation. States and regions sometimes offergovernment or not-for-profit options that canprovide this service. In New York State, forexample, the SUNY schools have access, for afee, to the Information Technology ExchangeCenter (ITEC), which supplies virtualizationhosting and other support.
Storage
The problem of meeting the increasingdemand for storage space can be compounded by different and competing needs for enterprise
and individual storage. A storage areanetwork (SAN) infrastructure can require significantcapital outlay, in an era of shorter and
shorter refresh cycles. With cloud-based storage,adding capacity is just a phone call ore-mail away, with options to customize solutionsto meet enterprise and individual needs.
Phasing in this service might start with aremote contract for SAN support. A next stepmight be to mirror your SAN infrastructure atan off-site location. As an institution migratesits server infrastructure to the cloud, someenterprise storage will likely migrate with it.Different services can then be contracted toprovide cloud-based solutions for individualstorage needs.
Examples of this service include NirvanixStorage Delivery Network, an appliance thatresides in an institution’s local data center.
On the back side, it attaches to variouscloud-based storage mechanisms. To theinstitution’s local infrastructure, it looks like a
large pool of local storage. To serve individualneeds, an institution can look at services likeDropBox and Box.net, which provide storagethat integrates into the user’s desktop. Forshort-term needs, Amazon offers the SimpleStorage Service (S3).
Backup
As the volume of data storage changes,backup tape or disk management can getcomplicated. The need to be compliant withe-discovery and retention laws, and the factthat much institutional data does not resideon central servers, can compound the issues.Moving backup to the cloud can simplifydata and compliance management and mitigatethe need to worry about tapes and disks.Cloud-based services can also include desktop/laptopbackup options, so an institutioncan be more confident about the ecoverabilityof that data.
Iron Mountain provides an enterprisebackup solution, and places like Mozy andCarbonite provide relatively inexpensive desktop and laptop backup options.
Desktops
One of the most significant challenges ofdesktop management is desktop applicationand patching. In addition, software in
computer labs on campus is not availableto off-site or distance learners. As desktoprefresh cycles move to three years or
less, pressures on both capital and staffingresources increase.
With cloud-based delivery, an institutioncan provide virtual services to local desktopsvia thin client or software client, with
only a few central images to manage. Withthis model, users can also access computerlab resources more easily when they are off campus.
Companies like Panologic and NComputingprovide hardware thin clients, while companieslike ICC Global and Secure-24 canprovide the back-end infrastructure for a virtualizeddesktop.
Telephony
Even the campus phone service is a candidatefor migration to the cloud. Manycampuses consider land lines a necessaryevil, but not necessarily a place forinvestment. Many institutions today run onunsupported analog PBX systems. Movinga telephony system to the cloud eliminatesthe need to manage a complicated PBX orVoIP system. The only things to manage on
premises are handsets.PAETEC provides a hosted IP telephonysolution. Apogeenet provides this service forstudents in residence halls.
Network
In essence, networks are the linchpin for allthe other services provided on campus, butthey are complicated and expensive to maintain.The increase of services on networksrequires increasingly specialized expertise.Networks, too, do not scale down well.
Moving this infrastructure to the cloudcan provide access to the skills and expertiseneeded to monitor and manage a
sophisticated network 7 × 24. It can betterscale both down and up, and it can stabilizecosts (albeit using operating dollars
instead of capital funds).Phasing-in of a cloud network mightstart by outsourcing resources for specific
network activities. The next step might beto partner with a firm to design, build, andmanage a network for you. A final phase
would be to purchase a network servicelevel from a provider. The network neverreally goes off campus, which will still house
hardware and wiring, but the operationaland potentially capital responsibilities shiftoff site.
Examples of network solutions includeNfrastructure, and Apogeenet for residential networks.
Final Considerations
Moving infrastructure to the cloud does notnecessarily save money. Money an institutionmight save by not investing in an individualserver could end up being spent on additionalbandwidth. The importance of IaaS isthat it allows institutions to hand off commodityor non-differentiating services so thatstaff can focus on things that do differentiate the institution.
In thinking about migrating to IaaS, campusIT administrators would do well to considerthree potential challenges:
Make the data center you have the lastone you will ever need.
Everything doesn’t need to be sourced tothe cloud at once, but start somewhere.
When opportunities arise, take them.
cloud computing infrastructure as a service pdf, infrastructure as a service iaas pdf, magic quadrant for cloud infrastructure as a service pdf
Source: http://docphy.com/technology/computers/software/infrastructure-as-a-service.html
No comments:
Post a Comment