AMERITAS INFORMATION TECHNOLOGY DISASTER RECOVERY AND DATA CENTER STRATEGY O VERVIEW There are currently 3 primary Data Center lcatins (Lincln, Cincinnati, and Calvert) and 2 secndary (Fallbrk and Philadelphia). In additin t these lcatins this strategy will als include Summit in Cincinnati, San Antni, and Wayne). The majrity f critical business applicatins cmprised f Plicy administratin, claim payment, and fund management reside n the mainframe lcated in the Lincln datacenter with web and nn-mainframe platfrms being deplyed in all facilities. STRATEGY 1. Migratin t a strategy f Active Disaster Recvery vs. Outage based Disaster Recvery. Future disaster recvery fr missin critical applicatins will be replaced with cntinuus peratins, aut failver, redundancy f data, and cntinuus backup/replicatin. 2. Prductin Applicatin Servers by default reside in Lincln (unless the criterin dictates therwise). 3. Disaster Recvery in any frm must cntain, at a minimum, tw gegraphically diverse facilities. Other Ameritas facilities shuld nt play an active rle in disaster recvery except where business critical envirnmental services must be maintained e.g. cre netwrking devices, cre telephny devices. The secndary r partner facility will be lcated in the Cincinnati area. 4. Technical skills t supprt servers can be lcated anywhere therefre the infrastructure may be deplyed anywhere. Depth f technical skills will be leveraged t prvide business cntinuity. 5. Clud cmputing is evaluated as new applicatins are purchased (e.g. Buy vs. Build) 6. Virtualizatin f Infrastructure everywhere pssible. 7. Minimize the number f secndary data centers. 8. Leveraging f hardened data centers either already existing as part f the Ameritas Cmpanies current envirnments r the lease f flr space in existing, third party hardened facilities. B ACKGROUND Disaster recvery has taken n many frms ver the years. Just as business needs have changed, s have business needs fr disaster recvery. In tday s business climate, simply being ffline fr an hur may cause significant impact t sales, reputatin, and service resulting financial lss as businesses becme increasingly dependent n real-time electrnic transactins. In the past, disaster recvery fcuses have changed as business evlutin has driven custmers and assciates t the web as a means f mitigating sftware deplyment, licensing, and availability challenges. A typical data center disaster recvery deplyment utilized ne facility that prvided all f the cmputing pwer and redundancies necessary t sustain the business. Disaster recvery was prvided by securing a facility, usually in a separate gegraphical lcatin, purchasing enugh hardware t prvide the level f cmputing pwer necessary t supprt the business until the cause f the disaster had been mitigated. This hardware sat idle and prvided n benefit t the business ther than supprt in a disaster and is generally expensive in that the business is paying t lease r share flr space, equipment, and external staff that nly prvides value in the event that the primary data center becmes unavailable. As cmpanies began t decentralize and grw thrugh M&A activities, multiple datacenters emerged but they still remained lsely cnnected t ne anther. Each datacenter tended t huse platfrms that serviced the ppulatin in clsest prximity and rarely did they serve as disaster recvery facilities
fr ne anther. Cmmnly, disaster recvery fllwed the standard f leasing flr space, equipment, and peratinal staff and tapes were still created and shipped t this external facility. Current trends are psitining businesses t be available at all times with regards t capability regardless f disaster. This mdel seeks t prvide the businesses that they serve with all necessary functinality t handle business transactins but at a reduced capacity. Advances in technlgy management, failver, and peratins have enabled this shift. Gartner has cnfirmed this in discussins with analysts and have identified upcming shifts t lcal data center deplyments within the next 10 years. BUSINESS DRIVERS Fully Supprt Business requirements Service Levels (availability, reliability, stability) Defined DR and BCP (business cntinuity planning) needs Lcal print (lwest cst f wnership) Affrdability CURRENT STATE P RIMARY D ATA C ENTER: LINCOLN, NE The primary Data Center is lcated in Lincln, NE and cntains the prductin mainframe, disk, tape and high speed Xerx laser printers. Additinally, servers that cntain critical business applicatins, including web servers, and file and print services reside in the primary Data Center. Since the merger, we have lcated mst new prductin servers in Lincln, NE unless there is capacity n existing hardware in the backup Data Center (i.e. initial build ut f BPM) that can be utilized and lwer expenses. The majrity f the Data Center envirnmentals have been updated recently. A new generatr backup was installed in 2008 prviding cntinuus peratins in the primary data center. The Pwer Distributin Units (PDU) and Air Cnditining units have been updated in the past 3 years. The UPS (battery backup) will need t be replaced in the next cuple f years. The primary telephne switch is lcated in Lincln, NE. with fail-ver (Enterprise Survivable Servers) units in Fallbrk, Cincinnati and San Antni. B ACKUP D ATA C ENTER: CINCINNATI, OH The backup Data Center in Cincinnati, OH has the disaster recvery mainframe and peripheral devices (disk and tape). Additinally, servers that cntain critical business applicatins, including web servers, and file and print services are lcated in the backup data centers. Tw high speed Xerx printers reside in the backup Data Center fr lcal printing. The decisin t identify a primary and backup Data Center during integratin mstly invlved bringing the mainframe platfrms tgether. Since that time sme server hardware has been refreshed in Cincinnati hwever mst new prjects that require hardware, the servers have been placed t Lincln. The data center has a generatr backup fr cntinuus peratins. The UPS batteries were replaced in January, 2010 and UPS will need t be replaced in the next few f years. D ISASTER R ECOVERY F ACILITY: FALLBROOK, NE
The Fallbrk lcatin cntains EMC strage and server hardware in the event f a disaster at O Street. This lcatin is als used t recver critical servers in the event f a disaster in Cincinnati. There is a space n the Grund Level that is dedicated t IT and disaster recvery. This was established premerger and is still maintained tday. The EMC DMX3000 Strage device was purchased in 2006 and data frm critical servers/applicatins is synchrnized with the disk strage in the primary Data Center utilizing the high speed fiber cnnectin. This allws fr Recvery Pint Objective f less than 5 minutes fr critical servers/applicatins. The mainframe data is synchrnized t Fallbrk and then cpied t the DMX4 strage device in Cincinnati. The hardware warranty n the EMC-3000 in Fallbrk expires in 2010. The Disaster Recvery strategy will drive the future rle f Fallbrk as it relates t a recvery site fr IT. The UPS batteries were replaced in February, 2010. The Calvert lcatin uses a traditinal, ffsite disaster recvery mdel where data is shipped peridically t a SunGard facility in Philadelphia. S TAFFING One f the stated gals f integratin was t have single management, multiple lcatins. T achieve this gal, a review f available remte management tls was perfrmed that has enabled staff t be lcated in bth lcatins regardless f where the hardware r sftware resides. As a result, we were able t utilize the vast knwledge already in huse, keep staff levels belw industry average, increase wrklads and nt reduce service levels. This was been valuable in the Infrastructure areas, including Operatins, Supprt Center, and Technical Services where acquiring institutinal knwledge can be very expensive. Currently, there is IT supprt staff in Lincln, NE and Cincinnati, OH. This includes all f Infrastructure (Technical Services, Netwrk, Telephny, Servers, Desktp, Operatins, Supprt Center, and Database Administratrs). Persnnel in each f these areas supprt platfrms and/r businesses acrss lcatins. Fr example, Cmputer Operatins has persnnel in each lcatin t perfrm specific tasks such as print and tape. Hwever, majrity f Cmputer Operatins supprt staff wh are c-lcated versee scheduling, execute batch jbs and supprt the applicatin teams fr all Ameritas Cmpanies activities. CRITERIA FOR DATA CENTER LOCATION The fllwing criteria will drive the strategy t manage technlgy assets the mst effective/efficient way fr the future Ameritas: Business criticality This criteria allws fr the establishment f hw critical an applicatin and/r prcess is t the wning business unit. Technical requirements This criteria is intended t meant t establish what requirements must be met fr the applicatin and/r prcess t functin frm a technical perspective. Staffing and supprt This criteria is meant t establish what staffing and supprt is necessary fr the applicatin and/r business prcess being evaluated. Infrastructure readiness (including envirnmentals) Is meant t define the ability f the infrastructure t supprt the strategy fr the specific applicatin and/r business prcess that is being evaluated as well as fr the entirety f the infrastructure t be able t supprt.
Prximity This criteria has tw measures. The first and primary is the lcatin f the applicatin and/r prcess t its architectural cmpnents. An example is the prximity f an applicatin t its required database. This secnd measure is an identificatin f the lcatin f the user ppulatin with respect t the applicatin and/r business prcess. Clser prximity t the applicatin may mean better user respnse time such as in a mainframe applicatin. Usage requirements This criteria is meant t establish hw frequently r heavily and applicatin and/r prcess is utilized. Higher and frequent usage tends t indicate higher business criticality and is has a relatinship t the prximity criteria. Cre vs. ther (apps, etc.) Meant t identify if the applicatin and/r prcess is cre t ur business. Similar t business criticality. Business cntinuity / DR / RPO & RTO Latency tlerance Meant t establish hw tlerant an applicatin and/r prcess is t netwrk induced latency. This can have impact n cmmits t a database, disk replicatin, and ther items. Business risk tlerance Meant t ascertain the level f risk a business is willing t assume fr the applicatin and/r prcess in questin. Size f lcatin Intended t place parameters arund the ability f a lcatin t supprt the needs f the strategy. NEAR- TERM! Evaluate current prductin applicatin servers nt currently in Lincln! Ratinalize what servers need t be relcated! Evaluate current infrastructure FUTURE STATE The Primary Data Center will cntinue t be lcated in Lincln, NE. with the mainframe and majrity f the prductin servers residing at this lcatin. High speed printing fr Lincln will be prduced and mailed frm this lcatin. The backup Data Center will be in Cincinnati, OH. with its primary use being disaster recvery fr all platfrms. There may be situatin where prductin servers may reside in the backup Data Center based n the strategy statements. High speed printing fr Cincinnati will be prduced and mailed frm this lcatin. Fallbrk will nt be used fr disaster recvery in the future. Critical business servers residing in the primary Data Center will be recvered at the backup Data Center. Any critical business servers residing in the backup Data Center will be recvered at the primary Data Center. This new strategy will prvide greater prtectin in the event f a reginal disaster in the Lincln, NE area. IT will als lwer verall csts as the third strage device will nt be required since replicatin f data will be between the tw Data Centers. The Calvert disaster recvery site, currently utsurced in Philadelphia, PA, will be in surced and mved t the backup Data Center in Cincinnati. This has the ptential t reduce csts ver 5 years and align the businesses in the Bethesda, MD area with the rest f the enterprise. This will als enable better asset management acrss the UNIFI Cmpanies.
Summit in Cincinnati, OH and Calvert s cre infrastructure shuld be architected and centrally lcated in the primary Data Center. S TAFFING With current technlgy available including mnitring tls, remte cnsles and remte access, lcatin f supprt persnnel in relatin t equipment is nt a significant factr in the decisin making prcess. As technical skills and institutinal knwledge becme mre valuable t Ameritas, we will cntinue t retain and recruit staff based n skills and nt lcatin, unless there is a specific business requirement t have staff at a certain lcatin. We will cntinue t manage staff levels belw industry averages and nt have duplicatin f effrts acrss lcatins. Furthermre, having staff in multiple lcatins prvides redundancy in the event f a reginal disaster.