2015 IOUG DATA PROTECTION AND AVAILABILITY SURVEY

Transcription

1 2015 IOUG DATA PROTECTION AND AVAILABILITY SURVEY By Joseph McKendrick, Research Analyst Produced by Unisphere Research, a Division of Information Today, Inc. September 2015 Sponsored by Produced by

2 2 TABLE OF CONTENTS Executive Summary 3 The State of Data Availability 4 The Continuing Scourge of Downtime 10 What the Business Wants 18 Taking Action Against Downtime 24 Demographics 30 IOUG Recommends 33

3 3 EXECUTIVE SUMMARY In today s fast-paced digital economy, no organization can afford latency in its delivery of information to decision makers or customers. Enterprises are under pressure to deliver services on a 24x7 basis, and their data has to be ready as well. However, while an impressive array of technology exists to deliver data on a real-time, continuous basis, even those organizations with multiple redundancies built into their systems and networks still face challenges keeping up. This is one of the key takeaways in a survey of 331 data managers and professionals, conducted by Unisphere Research, a division of Information Today, Inc., among data managers and professionals who are members of the Independent Oracle Users Group (IOUG). Respondents came from organizations of all sizes and across various industries. A demographic overview is available at the end of the report. The key findings of the study include the following: n Multiple data centers are the rule among most enterprises most have one or two data centers, while 30% have three or more. Most are configured to be on hot standby, and handle multiple database types. n Unplanned downtime has cost businesses over the past year in terms of lost productivity and a loss in customer confidence. On average, businesses forfeit one business day a year to downtime. One in seven also report severe data losses. n Businesses are seeking more uptime from service-level agreements, but IT departments are struggling to provide it. More than three in five data managers are not fully confident they can meet these SLAs in the event of a system disruption. Only 28% report that, at best, they are only sometimes meeting their agreed-upon SLAs for downtime. n One-third of respondents report low levels of satisfaction with their current enterprise data availability strategies. Most report difficulties with funding new efforts. Many are turning to cloud and virtualization. On the following pages are the detailed results and analysis of the study.

4 4 THE STATE OF DATA AVAILABILITY Multiple data centers are the rule among most enterprises most have one or two data centers, while 30% have three or more. Most are configured to be on hot standby, and handle multiple database types. Enterprises have a number of options for ensuring the rapid availability of applications, systems, and data, and have adopted many of these approaches. The three leading options identified among survey respondents include maintaining standby servers at a remote site, data snapshots, and storage virtualization. Interestingly, 40% still indicate they employ tape backup, with tapes being shipped offsite, an availability strategy used for decades that is hampered by a great deal of latency. (See Figure 1.) One of the most robust approaches to availability is to maintain or contract for a secondary data site, preferably some distance from the primary center. To develop a better posture to manage business continuity and data availability, multiple data centers are the rule among most enterprises. Most respondents have one or two data centers, while 30% have three or more. These numbers do not include enterprises using cloud services for backup purposes. (See Figure 2.) We will be contrasting the views of respondents with multiple data centers with single data center sites throughout this report. Hot standby is also the rule for many of today s multidata center sites. Among respondents with two or more fullfunctioning data centers, the leading type of configuration is an active-active setup, indicated by a total of 43% of respondents. The most common configurations include active-active sites, with the secondary site serving as a read-only mirrored site, as well as two active sites. Just under one-third of organizations maintain two fully active read-write sites, in which the secondary data site serves as a backup disaster recovery or continuity of business environment. (See Figure 3.) Respondents were asked whether they support data replication of their multiple databases between these data centers. Among those with two or more data centers, a majority replicate both Oracle and non-oracle production databases to one or more secondary sites. (See Figure 4.) Bearing in mind that the survey base consists of IOUG members, and thus tends to be Oraclecentric, it s interesting to note the heavy presence of non-oracle data environments in these arrangements as well. The non- Oracle databases most likely to be seen at secondary sites include Microsoft SQL Server (cited by 61%) and MySQL (24%). Onefifth also transmit data from IBM DB2 databases as well. (See Figure 5.) Data availability platforms are being used to support live production data flowing through enterprises, rather than data for analytical or secondary purposes. Among the data being backed up to these secondary sites, a majority of respondents, 53%, indicate that the bulk of the data they are backing up can be considered mission-critical enterprise data versus departmental, dev/test, user support, or infrastructure support. (See Figure 6.) The percentage of mission-critical data subject to back up and protection varies little, regardless of the sophistication of the availability environment. Fifty-three percent of respondents with three or more data centers say the bulk of their production data is backed up, compared to 50% of those with single data center environments. (See Figure 7.)

5 5 Figure 1: Availability Solutions and Strategies Standby servers on a remote site 46% Snapshots 46% Storage virtualization 46% Asynchronous replication 40% Tape backup and tapes sent offsite 40% Cloning 35% Backup appliance 35% Synchronous replication 34% Virtual servers with restart capabilities 32% (including cloud) Replication of both applications and 31% data (such as virtual machine images) Continuous data protection and 28% availability Active-active instances of applications 19% (including cloud) WAN-based replication of backup 19% and recovery images to second site Automatic backup to the cloud 16% Dynamic mobility of virtual applications 8% between data centers Don t know/unsure 7% Other 1% (Multiple responses permitted.

6 6 Figure 2: Number of Full-Functioning Data Centers None 4% One 21% Two 35% Three 12% Four 4% Five or more 8% Don t know/unsure 7% (Total does not equal 100% due to rounding.) Figure 3: Data Center Configurations (Among respondents with multiple data centers.) Active-passive; one primary with 30% secondary offline disaster recovery site(s) Active-active; one primary with 20% secondary mirrored site(s) Active-active; resources shared in real time 19% Active-active; one primary with secondary 5% cloud site(s) Don t know/unsure 18% Other 8%

7 7 Figure 4: Replicate Production Databases to Secondary Sites? (Among respondents with multiple data centers). Don t know/unsure 9% No 32% ORACLE Yes 59% Don t know/unsure 11% No 35% NON-ORACLE Yes 53% (Total does not equal 100% due to rounding.)

8 8 Figure 5: Non-Oracle Production Databases Replicated at Secondary Sites Microsoft SQL Server 61% MySQL 24% IBM DB2 19% NoSQL databases (any brand) 11% Sybase ASE 6% Don t know/unsure 22% Other 9% (Multiple responses permitted. Figure 6: Percentage of Mission-Critical Enterprise Production Data Backed Up to Secondary Sites 1% to 25% 13% 26% to 50% 26% 51% to 75% 23% 76% to 99% 16% 100% 14% Don t know/unsure 7% (Total does not equal 100% due to rounding.)

9 9 Figure 7: Percentage of Production Data Backed Up to Secondary Sites By Number of Data Centers One Two Three or more 1% to 25% 13% 12% 14% 26% to 50% 31% 31% 24% 51% to 75% 21% 26% 24% 76% to 99% 18% 15% 15% 100% 11% 14% 14%

10 10 THE CONTINUING SCOURGE OF DOWNTIME Unplanned downtime has cost businesses over the past year in terms of lost productivity and a loss in customer confidence. On average, businesses forfeit one business day a year to downtime. One in seven also report severe data losses. The purpose of data availability strategies, of course, is to eliminate, or at least mitigate, the possibilities of information going offline due to systems disruptions. While there has been a great deal of progress over the decades in managing the threats of downtime, there is still a high risk of data loss in the time between the occurrence of an incident and the restart of the backup system. For example, transactions taking place at the time of a disruption may be lost. Unplanned downtime can occur for any number of reasons from hackers violating a system, to errors in software updates, to natural disasters. These incidents, no matter how minor, carry a high cost. These costs are real to the business, and can be felt almost immediately. Loss of employee productivity because of interrupted work, or needing to wait while systems get back up and running is the leading effect. A majority of enterprises, 54%, say they have suffered productivity losses as a result of downtime. There is also an impact on the way customers perceive businesses as well. Onethird of respondents report that their organizations suffered a loss of customer confidence and loyalty in the past year when their systems went down. Fully 28% report direct loss of revenues, and a similar number say innovation has suffered, as downtime cut into product or service development. (See Figure 8.) Even with a secure business infrastructure, protected data and security-compliant end users, organizations simply cannot operate if their data or systems are unavailable for minutes, hours or days at a time, said a respondent. We protect by first guarding our data from disaster, then deploying solutions to keep your critical systems highly available to your employees, customers and other users. Unplanned downtime is the scenario that enterprises constantly prepare for and anticipate as a potential disruption. For purposes of this survey, unplanned downtime is defined as time during which a system cannot be used for normal productive operations due to unforeseen failure in hardware, software or database components, or human errors. There is a significant degree of accumulated unplanned downtime that enterprises in this survey have encountered with their mission-critical applications over the past year. Twenty-two percent report that the total length of their downtown over the past year totaled more than eight hours, or one entire business day. (See Figure 9.) The larger and more sophisticated the data availability environment, the more accumulated downtime for these systems over the past year. Twenty-six percent of the respondents with multi-data center environments (three or more) report accumulated downtime of more than one business day (eight hours), versus 16% of those with single data center sites. While multi-data center environments are comprised of many more systems and applications, and thus more subject to downtime, this also points to the need for more innovative approaches above and beyond scaling with hardware. (See Figure 10.) There is data loss that goes with downtime as well. At least 15% of respondents estimate that they have incurred significant data losses defined as more than one terabyte s worth of data over the past year due to unplanned downtime or other events. Close to one in 20 enterprises even report they saw losses of more than 10 terabytes worth of data just in one year s time. (See Figure 11.) Having redundant data centers does help stem data losses. Among larger sites with three or more data centers, a total of 15% report losses of greater than a terabyte of data over the past year compared to 17% of single data center sites. Considering the likely total sizes of these data centers and host companies, this suggests that single data center sites are taking a larger proportion of losses. (See Figure 12.) There are a myriad of causes of unplanned downtime. The type of incident that stands out among all others, however, is network outages, blamed by a majority of respondents for downtime over the past three years. Of course, this is clearly a situation that is out of the control of database managers and professionals, who nonetheless need to have workarounds ready when issues outside of their domains do occur. Faulty hardware also can bring data environments down, with close to two-fifths having suffered server failures in recent years. More than onethird report that human errors have brought their environments to a screeching halt. Interestingly, the thing that every data manager fears and prepares for a natural disaster is actually low on the list, having been the case of downtime in only 8% of enterprises in the last three years. (See Figure 13.) While many of these factors are out of the direct control of data managers and professionals, the onus is on them to keep information flowing at all times. Businesses are becoming increasingly sensitive to the disruptions that occur due to downtime. In an economy that requires being on and able to deliver services on a 24x7 basis, any latency that occurs may have a steep price. Most enterprises are pressing their IT and data management staffs to take measures to keep unplanned downtime to a minimum. A majority of respondents, 54%, indicate that their organizations have been, to some degree, mandating reductions in downtime over the past three years. Close to one-fourth say this pressure to curb downtime has been significant, suggesting that new methods and processes are called for. (See Figure 14.) The main reason businesses are pushing for reductions in downtime is because the 24x7 business rules. Seventy-one percent of respondents state that their businesses are calling for reduced downtime in order to support always-on availability t t

11 11 to customers and partners. Data availability itself is a concern for most businesses. In addition, half of the respondents indicate that real-time processing is now a priority for their businesses. Big data and mobile two emerging paradigms also are driving the concern about downtime among one-third of the businesses covered in the survey. (See Figure 15.) Planned downtime is another challenge in today s digital economy. While planned downtime (defined as when systems and databases are purposely taken offline by IT departments for maintenance, upgrades, or migrations) can be managed in most cases, such interruptions need to be carefully controlled to keep business disruption to a minimum. For example, many organizations perform system maintenance over weekends to avoid cutting into critical business hours. Still, in a 24x7 economy, there may be a penalty. Among survey respondents, at least 39% report suffering the loss of more than one business day due to planned downtime. (See Figure 16.) Systems upgrades and server maintenance have been the leading causes of planned downtime impacting mission-critical applications over the past three years, respondents report. More than two-thirds indicate they have had to take their systems down to perform upgrades. Another 60% say the leading cause of planned downtime is routine server maintenance. Database performance tune-ups or maintenance is the third major cause of planned downtime, cited by close to half of respondents. (See Figure 17.) To a large extent, there is complacency when it comes to dealing with unplanned downtime. Organizations do not have precise or overly ambitious goals for reducing annual planned downtime with mission-critical applications and databases over the next 24 months. At least 40% say that they have no goals, and that the current levels of availability are acceptable. Another 20% say their businesses seek reductions of up to one-quarter of their current downtime. (See Figure 18.) Figure 8: Business Impact of Data Loss/Downtime over Past 12 Months Loss of employee productivity 54% Loss of customer confidence/loyalty 33% Loss of revenue 28% Delay in product/service development 28% Delay in getting products/services 14% to market Loss of customers 10% Loss of a new business opportunity 10% Loss of business to a competitor 9% Loss of an incremental business opportunity 8% Loss of repeat business 8% There have been no business consequences 16% Don t know/unsure 15% Other 1% (Multiple responses permitted.)

12 12 Figure 9: Accumulated Unplanned Downtime With Mission-Critical Applications Over the Past Year <1 hour of unplanned downtime 23% 1 to 4 hours 28% 4 to 8 hours 15% 8 to 24 hours 4% >24 hours 8% Don t know/unsure 12% Figure 10: Accumulated Unplanned Downtime By Number of Data Centers One Two Three or more <1 hour of unplanned downtime 19% 29% 17% 1 to 4 hours 31% 29% 29% 4 to 8 hours 13% 15% 19% 8 to 24 hours 9% 14% 16% >24 hours 7% 6% 10%

13 13 Figure 11: Average Amount of Data Lost Over the Past Year Due to Unplanned Downtime or Other Events Don t know/unsure 32% <1TB 53% 10TB to 50TB 2% >50TB 2% 1TB to 10TB 11% Figure 12: Average Amount of Data Lost Over the Past Year By Number of Data Centers One Two Three or more <1TB 56% 56% 52% 1TB to 10TB 17% 10% 10% 10TB to 50TB 0% 2% 4% >50TB 0% 4% 1%

14 14 Figure 13: Leading Causes of Unplanned Downtime With Mission-Critical Applications Over Past Three Years Network outages 57% Server failures 38% Human errors 34% Application errors 30% Power outages 27% Storage failures 27% Complications from application or 18% infrastructure refreshes Usage spikes/surges 14% Third-party supplier or cloud outages 12% Data corruption 11% Natural disaster/weather event 8% Hackers/security breaches 6% Don t know/unsure 10% Other 1% (Multiple responses permitted.)

15 15 Figure 14: Organization Mandating Reductions in Downtime Over Past Three Years? Don t know/unsure 8% No change in 38% requirements Yes, signficantly 22% Yes, moderately 32% Figure 15: Trends Driving Business Demands for Reduced Downtime Need for 24x7 business availability 71% Maintaining data availability 54% Requirements for real-time processing 50% Managing increasingly large data volumes 34% Increasing use of mobile devices 32% Increasing query loads/usage spikes 29% Prioritization and type of data to be 22% made available Social media 16% Don t know/unsure 2% Other 1% (Multiple responses permitted.)

16 16 Figure 16: Accumulated Planned Downtime with Mission-Critical Applications Over the Past Year <1 hour of unplanned downtime 10% 1 to 4 hours 22% 4 to 8 hours 17% 8 to 24 hours 22% >24 hours 17% Don t know/unsure 12% Figure 17: Leading Causes of Planned Downtime with Mission-Critical Applications Over the Past Three Years System upgrades 67% Server maintenance 60% Database performance and maintenance 47% Database or application migrations 45% Hardware migrations 38% Application migrations 29% Bringing up new applications 19% Testing and QA 13% Data replication/mirroring 10% Cutting over to a new data center 11% Managing external data 6% Don t know/unsure 6% Other 3% (Multiple responses permitted.)

17 17 Figure 18: Business Goals for Reducing Annual Planned Downtime With Mission-Critical Applications and Databases Over Next 24 Months None current level is acceptable 40% Up to 25% reduction 20% Between 25% to 75% 10% Between 50% to 75% 4% Between 75% to 100% 5% Don t know/unsure 21%

18 18 WHAT THE BUSINESS WANTS Businesses are seeking more uptime from service-level agreements, but IT departments are struggling to provide it. More than three in five data managers are not fully confident they can meet these SLAS in the event of a system disruption. Only 28% report that, at best, they are only sometimes meeting their agreed-upon SLAs for downtime. Not too long ago, businesses could be highly selective in deciding what data needs to be made available right away, versus data that could wait an hour or two, or even days. Now, however, in an always-on economy, data from across enterprises from inventory to finance to supply chain to human resources needs to be readily accessible, in real time or near real time. At least 43% of respondents indicate that the majority of their enterprise data currently needs to have high availability. (See Figure 19.) Today applications capture a huge amount of data terabytes and petabytes of data per second which is stored in the database, said a respondent. Because of that, database availability is a critical thing. We distribute, replicate, and back up frequently to avoid any delay or loss of data transmitted. When working with the business to determine data availability requirements, the need for high availability is reflected in the service-level agreements (SLAs) drawn up. About one-third of respondents now seek less than one hour of downtime per year, as reflected in SLAs reported by respondents. For another 25%, one business day (eight hours) is the maximum amount of downtime their businesses will tolerate. (See Figure 20.) Meeting SLAs for downtime is an issue for more than one in four enterprises. A total of 28% of data managers and professionals report that, at best, they are only sometimes meeting their agreed-upon SLAs for downtime. This is even more problematic when multiple data centers are involved, with more moving parts that could go down. (See Figure 21.) Accordingly, 63% of data managers and professionals indicate they do not have full confidence that, in the event of a data loss incident, they can fully recover systems and data today from all platforms (on-premises and off-premises), in order to meet these SLAs. (See Figure 22.) Confidence in the ability to meet downtime SLAs doesn t necessarily increase with the number of data centers deployed. Among respondents in multiple data center environments, only 36% express extreme confidence compared to 42% of those with only one data center. (See Figure 23.) For most enterprises, mission-critical enterprise applications need to be up and running within two hours. A majority of respondents, 56%, indicate that their recovery time objectives (RTOs) are two hours or less. (See Figure 24.) In most cases, these RTOs are being met. When asked how long it actually took to recover and restore a mission-critical application following their most recent unplanned outage, 52% stated they were back up within two hours. (See Figure 25.) When it comes to recovering and getting back to business, multiple data centers makes the difference. Twenty-nine percent of respondents with three or more data centers indicate they were back up and running in less than 15 minutes, compared to only six percent of sites with single data centers. (See Figure 26.) For a majority of respondents, their recovery point objectives (RPOs) for their most important mission-critical enterprise applications need to capture any transactions and data that were happening right up until the instant a disruption occurs. Fiftynine percent indicate their RPO requirement is any point in time up to the point of failure. (See Figure 27.) There are a number of business goals for respondents RPOs for their most important mission-critical enterprise applications. A majority, 51%, seek to recover to any point in time needed (operational recovery), while 42% say they intend not to lose any data or transactions. Another 42% would be satisfied with minimal transaction loss. Only 27% would tolerate a minimal data loss. (See Figure 28.)

19 19 Figure 19: Percentage of Enterprise Data Needing High Availability None 2% 1% to 25% 20% 26% to 50% 27% 51% to 75% 18% 75% to 99% 17% 100% 8% Don t know/unsure 9% Figure 20: Business Service-Level Agreement Categories for Downtime <99.9% availability 33% (>8 hours, 45 minutes downtime/year) All others 11% 99.99% availability 16% (<52 minutes downtime/year) 99.9% availability 25% (<8 hours, 45 minutes downtime/year) % availability 15% (<5 minutes downtime/year)

20 20 Figure 21: Currently Meeting Service-Level Agreements for Downtime in Data Centers? In At Least 1 Data Center Across Multiple Data Centers Yes 62% 48% Sometimes 22% 22% Not frequently enough 5% 6% Not at all 1% 3% Don t know/unsure 6% 14% N/A 3% 7% (Total does not equal 100% due to rounding.) Figure 22: Confidence in Meeting Downtime SLAs Very confident 36% Not at all confident 1% Not very confident 3% Moderately confident 47% Some doubt 12% (Total does not equal 100% due to rounding.)

21 21 Figure 23: Confidence in Meeting Downtime SLAs By Number of Data Centers One Two Three or more Very confident 42% 34% 36% Moderately confident 40% 49% 47% Some doubt 14% 11% 13% Not very confident 4% 2% 4% Not at all confident 0% 3% 0% Figure 24: Recovery Time Objectives for Most Important Mission-Critical Enterprise Applications <15 minutes 19% 15 to 60 minutes 21% 1 to 2 hours 16% 2 to 4 hours 13% 4 to 8 hours 10% 8 to 24 hours 5% >24 hours 3% Don t know/unsure 12% (Total does not equal 100% due to rounding.)

22 22 Figure 25: Actual Recovery Times During Most Recent Unplanned Outages of Mission Critical Applications <15 minutes 19% 15 to 60 minutes 26% 1 to 2 hours 17% 2 to 4 hours 12% 4 to 8 hours 7% 8 to 24 hours 3% >24 hours 1% Don t know/unsure 14% Figure 26: Enterprises Recovering in Less Than 15 Minutes By Number of Data Centers 30% 25% 20% 15% 10% Two 19% Three or more 29% 5% 0% One 6%

23 23 Figure 27: Recovery Point Objectives for Most Important Mission-Critical Enterprise Applications Any point in time up to the point of failure 59% Any daily backup in the past 1 2 weeks 13% Any daily backup in the past month 4% Any weekly backup in the past 1 2 months 3% Any weekly backup in the past three months 2% Any monthly backup in the past year 1% Don t know/unsure 14% Other 4% Figure 28: Business Goals of RPOs Recovering to any point in time needed 51% (operational recovery) No data lost 42% No transactions lost 42% Minimal transactions loss 42% Minimal data loss 27% Recovery of historical data prior to failure 25% Recovery of incoming transactions 15% post-failure N/A 1% Don t know/unsure 12% (Multiple responses permitted.)

24 24 TAKING ACTION AGAINST DOWNTIME One-third of respondents report low levels of satisfaction with their current enterprise data availability strategies. Most report difficulties with funding new efforts. Many are turning to cloud and virtualization. Data availability is an area in need of rethinking. While these approaches are functioning and working, they are not inspiring a great deal of confidence among data managers and professionals. One-third of respondents report low levels of satisfaction with their current enterprise data availability strategies. This signals that it s time for a new approach to this problem. (See Figure 29.) Having a robust and elaborate data center infrastructure does not seem to alleviate this sense of dissatisfaction with the ability to keep data highly available for the business. The percentage of respondents with three of more data centers expressing that they are satisfied to very satisfied totals 65%, only a notch above 63% of respondents with a single data center. (See Figure 30.) One respondent described the ability to deliver availability to the business under tight constraints as the right balance of efficiency and effectiveness. Maintaining and adapting a highperformance data center is becoming increasingly complex. The inter-dependencies of our systems servers, storage and network can create a domino effect when even minor changes are made, the data manager said. What s hampering efforts to attain greater levels of data availability delivering information quickly, when and where it is needed, regardless of any underlying systems issues? The leading roadblock, cited by 60% of respondents, is organizational support. Additional data centers, tools, and people to run these processes, can be expensive. Data managers say a lack of budget or funding holds up efforts to improve the process. This is where education is often required, one respondent pointed out. We let executive personnel know how important it is to consider and have a data protection strategy and solution in order to guarantee the business that there will not be any loss of data. By building business cases that show the monetary loss and credibility loss of customers, they understand that we need to invest in technology solutions. A secondary, more technical problem is the performance drags that are created as a result of ongoing backup and replication. (See Figure 31.) While pressing business requirements in the always-on economy require new approaches to data availability, many enterprises are approaching the problem with hardware and systems upgrades. A majority state their preferred action to address data availability issues over the past three years has been to purchase or upgrade their systems and hardware. Still, close to half recognize the need to increase the flexibility of movement of data across environments, and have made the move to data virtualization. About one in four also indicate they are adopting cloud services as an approach to ensuring the availability of their data. (See Figure 32.) More than one-third of enterprises, 34% indicate that a majority of their data is still backed up using traditional tape backup methodology. Just over one-quarter of respondents, 27%, report that the majority of their enterprise data is protected with snapshots. More than one-third of respondents, 38%, report that the majority of their enterprise data is protected with replication. (See Figure 33.) Private or hybrid clouds are the most likely forms of data protection cloud services to be leveraged within the next 12 months, the survey finds. Close to one-third report using these environments. (See Figure 34.) Close to half of respondents, 45%, report they have virtualized their Oracle environments. (See Figure 35.) How much of respondents enterprise mission-critical data are part of their virtualized environments? Close to two-thirds, 62%, report that more than half their mission-critical data is included in their virtualized environments. (See Figure 36.) Many companies are experienced with virtualization. Half of respondents indicate they now have had virtualization in place three or more years at their enterprises. (See Figure 37.)

25 25 Figure 29: Satisfaction With Current Enterprise Data Availability Strategies Very satisfied 25% Satisfied 37% It s adequate 27% Not satisfied 6% Very unsatisfied 2% Don t know/unsure 2% (Total does not equal 100% due to rounding.) Figure 30: Satisfaction with Data Availability Strategies By Number of Data Centers One Two Three or more Very satisfied 20% 21% 29% Satisfied 43% 38% 36% It s adequate 24% 30% 24% Not satisfied/very unsatisfied 12% 10% 8%

26 26 Figure 31: Greatest Challenges for Planning and Managing High-Availability Data Strategy Lack of budget/funding 60% Performance concerns 33% Configuration issues 25% Maintaining two or more separate 23% data sites Ensuring consistency with other 21% mission-critical applications Lack of overall strategy 21% Managing availability across multiple 21% databases: Meeting recovery time objectives (RTOs) 20% Issues with distances between primary 19% and backup sites Keeping track of all data sources 16% Other 20% (Multiple responses permitted.)

27 27 Figure 32: Steps Taken to Address Database Availability Issues Over Past Three Years Purchased and upgraded or replaced 52% systems/hardware Data virtualization 46% Data center modernization 46% Instituted new or updated policies and 35% procedures Implemented or subscribed to cloud-based 23% services Built/opened additional data center(s) 17% No specific actions taken during this time 13% Don t know/unsure 9% Other 2% (Multiple responses permitted.) Figure 33: Amount of Enterprise Data Backed Up Using Key Solutions Tape Snapshots Replication None 25% 15% 8% 1% to 25% 17% 24% 25% 26% to 50% 14% 14% 14% 51% to 75% 10% 12% 18% 75% to 99% 12% 9% 11% 100% 12% 6% 9% Don t know/unsure 11% 20% 14%

28 28 Figure 34: Data Protection Cloud Services Used or To Be Adopted in Next 12 Months Private or hybrid cloud 31% Backup-as-a-service 20% Archive-as-a-service 12% Data recovery-as-a-service 11% None of the above 34% Don t know/unsure 28% (Multiple responses permitted.) Figure 35: Virtualized Oracle Environments? No 21% Yes 45% Don t know/unsure 9% (Total doest not equal 100% due to rounding.) Under consideration/planning 24%

29 29 Figure 36: Percentage of Enterprise Mission-Critical Data in Virtualized Environments 1% to 25% 19% 26% to 50% 9% 51% to 75% 14% 76% to 99% 30% 100% 18% Don t know/unsure 9% (Total does not equal 100% due to rounding.) Figure 37: Length of Time Virtualization Has Been in Place 1 to 2 years 30% <1 year 12% Don t know/unsure 7% 3 to 4 years 28% 5 years or more 22% (Total doest not equal 100% due to rounding.)

30 30 DEMOGRAPHICS Figure 38: Respondents Primary Job Titles Database administrator 43% Director/manager of IS/IT or 13% computer-related function Chief information officer/chief technology 7% officer/vice president IT IT operations manager 6% IT consultant for IT service/integration firm 5% IT consultant independent contractor 5% Systems administrator 4% Analyst/systems analyst 4% Applications administrator 3% Programmer/developer 2% Manager of a business unit 2% Executive management level for the business 1% Other 5%

31 31 Figure 39: Number of Databases Running (Includes production, development and testing databases.) <10 15% 11 to % 101 to 1,000 35% 1001 to 5,000 10% 5,001 or more 5% (Total does not equal 100% due to rounding.) Figure 40: Number of Employees in Respondents Organizations (Includes all locations, branches, and subsidiaries.) 1 to 1,000 34% 1,001 to 5,000 22% 5,001-to 10,000 12% >10,000 27% Don t know/unsure/na 4% (Total does not equal 100% due to rounding.)

32 32 Figure 41: Respondents Primary Industries IT services/consulting/system integrator 18% Government (all levels) 16% Financial services 9% Education (all levels) 8% Healthcare/medical/life sciences 7% Computer software 7% Manufacturing 7% Business services 5% Insurance 4% Retail/distribution 3% Telecommunications/media 3% Consumer services 2% Computer hardware 2% Transportation 2% Utility 2% Other 3% (Total does not equal 100% due to rounding.)

33 33 IOUG RECOMMENDS This survey of 331 data executives and professionals finds that achieving 24x7 availability requires a holistic strategy and commitment from business and IT leaders. Enterprises seeking to improve their data and application availability need to consider the following strategies: Work closely with the business to determine where availability investments should be made. The business has to determine which datasets and applications need to be highly available, and investments should be targeted at these elements. Don t rely on a single point solution. Achieving always-on availability requires a set of solutions that address continuity and data protection across all key layers of the infrastructure: at the application level, host level, storage level as well as the data center level. Move to virtualization and cloud. Virtualization and cloud computing offer new approaches to data and application availability without the need to build out infrastructure. There are a variety of options, including providing backup through a cloud service, or using an internal or hybrid cloud solution. Manage expectations. Work with the business to set realistic thresholds with their service level agreements (SLAs) that are realistic and attainable within the scope and limitations of budgets and organizational commitment.