The Experts below are selected from a list of 198 Experts worldwide ranked by ideXlab platform

Dl Sisterson - One of the best experts on this subject based on the ideXlab platform.

  • Atmospheric Radiation Measurement Program Climate Research Facility Operations Quarterly Report: October 1 - December 31, 2010
    2011
    Co-Authors: Dl Sisterson
    Abstract:

    Individual raw datastreams from instrumentation at the Atmospheric Radiation Measurement (ARM) Climate Research Facility fixed and mobile sites are collected and sent to the Data Management Facility (DMF) at Pacific Northwest National Laboratory (PNNL) for processing in near-real time. Raw and processed data are then sent approximately daily to the ARM Archive, where they are made available to users. For each instrument, we calculate the ratio of the actual number of processed data records received daily at the Archive to the expected number of data records. The results are tabulated by (1) individual datastream, site, and month for the current year and (2) site and fiscal year (FY) dating back to 1998. The U.S. Department of Energy (DOE) requires national user facilities to report time-based operating data. The requirements concern the actual hours of operation (ACTUAL); the estimated maximum operation or uptime goal (OPSMAX), which accounts for Planned Downtime; and the VARIANCE [1 - (ACTUAL/OPSMAX)], which accounts for unPlanned Downtime. The OPSMAX time for the first quarter of FY2010 for the Southern Great Plains (SGP) site is 2097.60 hours (0.95 x 2208 hours this quarter). The OPSMAX for the North Slope Alaska (NSA) locale is 1987.20 hours (0.90 x 2208)more » and for the Tropical Western Pacific (TWP) locale is 1876.80 hours (0.85 x 2208). The first ARM Mobile Facility (AMF1) deployment in Graciosa Island, the Azores, Portugal, continued through this quarter, so the OPSMAX time this quarter is 2097.60 hours (0.95 x 2208). The second ARM Mobile Facility (AMF2) began deployment this quarter to Steamboat Springs, Colorado. The experiment officially began November 15, but most of the instruments were up and running by November 1. Therefore, the OPSMAX time for the AMF2 was 1390.80 hours (.95 x 1464 hours) for November and December (61 days). The differences in OPSMAX performance reflect the complexity of local logistics and the frequency of extreme weather events. It is impractical to measure OPSMAX for each instrument or datastream. Data availability reported here refers to the average of the individual, continuous datastreams that have been received by the Archive. Data not at the Archive are caused by Downtime (scheduled or unPlanned) of the individual instruments. Therefore, data availability is directly related to individual instrument uptime. Thus, the average percentage of data in the Archive represents the average percentage of the time (24 hours per day, 92 days for this quarter) the instruments were operating this quarter. Summary. Table 1 shows the accumulated maximum operation time (Planned uptime), actual hours of operation, and variance (unPlanned Downtime) for the period October 1-December 31, 2010, for the fixed sites. Because the AMFs operate episodically, the AMF statistics are reported separately and not included in the aggregate average with the fixed sites. This first quarter comprises a total of 2,208 possible hours for the fixed sites and the AMF1 and 1,464 possible hours for the AMF2. The average of the fixed sites exceeded our goal this quarter. The AMF1 has essentially completed its mission and is shutting down to pack up for its next deployment to India. Although all the raw data from the operational instruments are in the Archive for the AMF2, only the processed data are tabulated. Approximately half of the AMF2 instruments have data that was fully processed, resulting in the 46% of all possible data made available to users through the Archive for this first quarter. Typically, raw data is not made available to users unless specifically requested.« less

  • Atmospheric Radiation Measurement Program Climate Research Facility Operation quarterly report July 1 - September 30, 2010.
    2010
    Co-Authors: Dl Sisterson
    Abstract:

    Individual raw datastreams from instrumentation at the Atmospheric Radiation Measurement (ARM) Climate Research Facility fixed and mobile sites are collected and sent to the Data Management Facility (DMF) at Pacific Northwest National Laboratory (PNNL) for processing in near real-time. Raw and processed data are then sent approximately daily to the ARM Archive, where they are made available to users. For each instrument, we calculate the ratio of the actual number of data records received daily at the Archive to the expected number of data records. The results are tabulated by (1) individual datastream, site, and month for the current year and (2) site and fiscal year (FY) dating back to 1998. The U.S. Department of Energy (DOE) requires national user facilities to report time-based operating data. The requirements concern the actual hours of operation (ACTUAL); the estimated maximum operation or uptime goal (OPSMAX), which accounts for Planned Downtime; and the VARIANCE [1-(ACTUAL/OPSMAX)], which accounts for unPlanned Downtime. The OPSMAX time for the fourth quarter of FY2010 for the Southern Great Plains (SGP) site is 2097.60 hours (0.95 2208 hours this quarter). The OPSMAX for the North Slope of Alaska (NSA) locale is 1987.20 hours (0.90 2208) and for the Tropicalmore » Western Pacific (TWP) locale is 1876.80 hours (0.85 2208). The first ARM Mobile Facility (AMF1) deployment in Graciosa Island, the Azores, Portugal, continues, so the OPSMAX time this quarter is 2097.60 hours (0.95 x 2208). The differences in OPSMAX performance reflect the complexity of local logistics and the frequency of extreme weather events. It is impractical to measure OPSMAX for each instrument or datastream. Data availability reported here refers to the average of the individual, continuous datastreams that have been received by the Archive. Data not at the Archive are caused by Downtime (scheduled or unPlanned) of the individual instruments. Therefore, data availability is directly related to individual instrument uptime. Thus, the average percentage of data in the Archive represents the average percentage of the time (24 hours per day, 92 days for this quarter) that the instruments were operating this quarter. Table 1 shows the accumulated maximum operation time (Planned uptime), actual hours of operation, and variance (unPlanned Downtime) for the period July 1-September 30, 2010, for the fixed sites. Because the AMF operates episodically, the AMF statistics are reported separately and not included in the aggregate average with the fixed sites. This fourth quarter comprises a total of 2208 possible hours for the fixed and mobile sites. The average of the fixed sites exceeded our goal this quarter. The Site Access Request System is a web-based database used to track visitors to the fixed and mobile sites, all of which have facilities that can be visited. The NSA locale has the Barrow and Atqasuk sites. The SGP site has historically had a Central Facility, 23 extended facilities, 4 boundary facilities, and 3 intermediate facilities. Beginning in the second quarter of FY2010, the SGP began a transition to a smaller footprint (150 km x 150 km) by rearranging the original instrumentation and new instrumentation made available through the American Recovery and Reinvestment Act of 2009 (ARRA). The Central Facility and 4 extended facilities will remain, but there will be up to 12 new surface characterization facilities, 4 radar facilities, and 3 profiler facilities sited in the smaller domain. This new configuration will provide observations at scales more appropriate to current and future climate models. The transition to the smaller footprint is ongoing through this quarter. The TWP locale has the Manus, Nauru, and Darwin sites. These sites will also have expanded measurement capabilities with the addition of new instrumentation made available through ARRA funds. It is anticipated that the new instrumentation at all the fixed sites will be in place by the end of calendar year 2011. AMF1 continues its 20-month deployment in Graciosa Island, the Azores, Portugal, that began on May 1, 2009. The AMF will also have additional observational capabilities by the end of 2011. The second ARM Mobile Facility (AMF2) was deployed this quarter to Steamboat Springs, Colorado, in support of the Storm Peak Lab Cloud Property Validation Experiment (STORMVEX). The first field deployment of the second ARM Mobile Facility will be used to validate ARM-developed algorithms that convert the remote sensing measurements to cloud properties for liquid and mixed phase clouds. Although AMF2 is being set up this quarter, the official start date of the field campaign is not until November 1, 2010. This quarterly report provides the cumulative numbers of scientific user accounts by site for the period October 1, 2009-September 30, 2010.« less

  • Atmospheric Radiation Measurement Program Climate Research Facility Operations Quarterly Report January 1?March 31, 2010
    2010
    Co-Authors: Dl Sisterson
    Abstract:

    The U.S. Department of Energy (DOE) requires national user facilities to report time-based operating data. The requirements concern the actual hours of operation (ACTUAL); the estimated maximum operation or uptime goal (OPSMAX), which accounts for Planned Downtime; and the VARIANCE [1 – (ACTUAL/OPSMAX)], which accounts for unPlanned Downtime

  • Atmospheric Radiation Measurement program climate research facilities quarterly report April 1 - June 30, 2009.
    2009
    Co-Authors: Dl Sisterson
    Abstract:

    Individual raw data streams from instrumentation at the Atmospheric Radiation Measurement (ARM) Program Climate Research Facility (ACRF) fixed and mobile sites are collected and sent to the Data Management Facility (DMF) at Pacific Northwest National Laboratory (PNNL) for processing in near-real time. Raw and processed data are then sent approximately daily to the ACRF Archive, where they are made available to users. For each instrument, we calculate the ratio of the actual number of data records received daily at the archive to the expected number of data records. The results are tabulated by (1) individual data stream, site, and month for the current year and (2) site and fiscal year (FY) dating back to 1998. The U.S. Department of Energy (DOE) requires national user facilities to report time-based operating data. The requirements concern the actual hours of operation (ACTUAL); the estimated maximum operation or uptime goal (OPSMAX), which accounts for Planned Downtime; and the VARIANCE [1 - (ACTUAL/OPSMAX)], which accounts for unPlanned Downtime. The OPSMAX time for the third quarter of FY 2009 for the Southern Great Plains (SGP) site is 2,074.80 hours (0.95 x 2,184 hours this quarter); for the North Slope Alaska (NSA) locale it is 1,965.60 hoursmore » (0.90 x 2,184); and for the Tropical Western Pacific (TWP) locale it is 1,856.40 hours (0.85 x 2,184). The ARM Mobile Facility (AMF) was officially operational May 1 in Graciosa Island, the Azores, Portugal, so the OPSMAX time this quarter is 1390.80 hours (0.95 x 1464). The differences in OPSMAX performance reflect the complexity of local logistics and the frequency of extreme weather events. It is impractical to measure OPSMAX for each instrument or data stream. Data availability reported here refers to the average of the individual, continuous data streams that have been received by the Archive. Data not at the Archive are caused by Downtime (scheduled or unPlanned) of the individual instruments. Therefore, data availability is directly related to individual instrument uptime. Thus, the average percentage of data in the Archive represents the average percentage of the time (24 hours per day, 91 days for this quarter) the instruments were operating this quarter. Table 1 shows the accumulated maximum operation time (Planned uptime), actual hours of operation, and variance (unPlanned Downtime) for April 1 - June 30, 2009, for the fixed sites. Because the AMF operates episodically, the AMF statistics are reported separately and are not included in the aggregate average with the fixed sites. The AMF statistics for this reporting period were not available at the time of this report. The third quarter comprises a total of 2,184 hours for the fixed sites. The average well exceeded our goal this quarter.« less

  • Atmospheric Radiation Measurement program climate research facility operations quarterly report January 1 - March 31, 2009.
    2009
    Co-Authors: Dl Sisterson
    Abstract:

    Individual raw data streams from instrumentation at the Atmospheric Radiation Measurement (ARM) Program Climate Research Facility (ACRF) fixed and mobile sites are collected and sent to the Data Management Facility (DMF) at Pacific Northwest National Laboratory (PNNL) for processing in near real-time. Raw and processed data are then sent daily to the ACRF Archive, where they are made available to users. For each instrument, we calculate the ratio of the actual number of data records received daily at the Archive to the expected number of data records. The results are tabulated by (1) individual data stream, site, and month for the current year and (2) site and fiscal year (FY) dating back to 1998. The U.S. Department of Energy (DOE) requires national user facilities to report time-based operating data. The requirements concern the actual hours of operation (ACTUAL); the estimated maximum operation or uptime goal (OPSMAX), which accounts for Planned Downtime; and the VARIANCE [1 - (ACTUAL/OPSMAX)], which accounts for unPlanned Downtime. The OPSMAX time for the second quarter of FY 2009 for the Southern Great Plains (SGP) site is 2,052.00 hours (0.95 x 2,160 hours this quarter). The OPSMAX for the North Slope Alaska (NSA) locale is 1,944.00 hours (0.90 x 2,160), and for the Tropical Western Pacific (TWP) locale is 1,836.00 hours (0.85 x 2,160). The OPSMAX time for the ARM Mobile Facility (AMF) is not reported this quarter because not all of the metadata have been acquired that are used to generate this metric. The differences in OPSMAX performance reflect the complexity of local logistics and the frequency of extreme weather events. It is impractical to measure OPSMAX for each instrument or data stream. Data availability reported here refers to the average of the individual, continuous data streams that have been received by the Archive. Data not at the Archive are caused by Downtime (scheduled or unPlanned) of the individual instruments. Therefore, data availability is directly related to individual instrument uptime. Thus, the average percentage of data in the Archive represents the average percentage of the time (24 hours per day, 90 days for this quarter) the instruments were operating this quarter. Summary. Table 1 shows the accumulated maximum operation time (Planned uptime), actual hours of operation, and variance (unPlanned Downtime) for the period January 1 - March 31, 2009, for the fixed sites. The AMF has completed its mission in China but not all of the data can be released to the public at the time of this report. The second quarter comprises a total of 2,160 hours. The average exceeded our goal this quarter

S Nandi - One of the best experts on this subject based on the ideXlab platform.

  • ICDE - Taming the Downtime: high availability in Sybase ASE 12
    Proceedings of 16th International Conference on Data Engineering (Cat. No.00CB37073), 2000
    Co-Authors: Sumita Raghuram, Surendra Ranganath, S. Olson, S Nandi
    Abstract:

    The new companion architecture in Sybase Adaptive Server Enterprise (ASE) 12 for high availability is supported on a 2-node cluster with each node running a separate ASE 12 server in a companion configuration. This architecture is designed to withstand a single point of failure for unPlanned outages, and allow both nodes to be used for productive workload during normal operation. It enables fast failover and data recovery, supports automatic client migration during failure, and integrates seamlessly with adjoining layers in multi-tier architecture. It supports single system presentation of data for applications, and presents a rich set of features/infrastructure to reduce the Planned Downtime. During failover and failback, only the persistent data component is moved between the companion ASEs, making it fast and efficient. Introducing the proxy databases, this architecture enables user databases to be visible and accessible from either of the companions by shipping the queries to the appropriate node and returning the results to the client.

  • ICDE - Taming the Downtime: high availability in Sybase ASE 12
    Proceedings of 16th International Conference on Data Engineering (Cat. No.00CB37073), 2000
    Co-Authors: Sumita Raghuram, Surendra Ranganath, S. Olson, S Nandi
    Abstract:

    The new companion architecture in Sybase Adaptive Server Enterprise (ASE) 12 for high availability is supported on a 2-node cluster with each node running a separate ASE 12 server in a companion configuration. This architecture is designed to withstand a single point of failure for unPlanned outages, and allow both nodes to be used for productive workload during normal operation. It enables fast failover and data recovery, supports automatic client migration during failure, and integrates seamlessly with adjoining layers in multi-tier architecture. It supports single system presentation of data for applications, and presents a rich set of features/infrastructure to reduce the Planned Downtime. During failover and failback, only the persistent data component is moved between the companion ASEs, making it fast and efficient. Introducing the proxy databases, this architecture enables user databases to be visible and accessible from either of the companions by shipping the queries to the appropriate node and returning the results to the client.

Sumita Raghuram - One of the best experts on this subject based on the ideXlab platform.

  • ICDE - Taming the Downtime: high availability in Sybase ASE 12
    Proceedings of 16th International Conference on Data Engineering (Cat. No.00CB37073), 2000
    Co-Authors: Sumita Raghuram, Surendra Ranganath, S. Olson, S Nandi
    Abstract:

    The new companion architecture in Sybase Adaptive Server Enterprise (ASE) 12 for high availability is supported on a 2-node cluster with each node running a separate ASE 12 server in a companion configuration. This architecture is designed to withstand a single point of failure for unPlanned outages, and allow both nodes to be used for productive workload during normal operation. It enables fast failover and data recovery, supports automatic client migration during failure, and integrates seamlessly with adjoining layers in multi-tier architecture. It supports single system presentation of data for applications, and presents a rich set of features/infrastructure to reduce the Planned Downtime. During failover and failback, only the persistent data component is moved between the companion ASEs, making it fast and efficient. Introducing the proxy databases, this architecture enables user databases to be visible and accessible from either of the companions by shipping the queries to the appropriate node and returning the results to the client.

  • ICDE - Taming the Downtime: high availability in Sybase ASE 12
    Proceedings of 16th International Conference on Data Engineering (Cat. No.00CB37073), 2000
    Co-Authors: Sumita Raghuram, Surendra Ranganath, S. Olson, S Nandi
    Abstract:

    The new companion architecture in Sybase Adaptive Server Enterprise (ASE) 12 for high availability is supported on a 2-node cluster with each node running a separate ASE 12 server in a companion configuration. This architecture is designed to withstand a single point of failure for unPlanned outages, and allow both nodes to be used for productive workload during normal operation. It enables fast failover and data recovery, supports automatic client migration during failure, and integrates seamlessly with adjoining layers in multi-tier architecture. It supports single system presentation of data for applications, and presents a rich set of features/infrastructure to reduce the Planned Downtime. During failover and failback, only the persistent data component is moved between the companion ASEs, making it fast and efficient. Introducing the proxy databases, this architecture enables user databases to be visible and accessible from either of the companions by shipping the queries to the appropriate node and returning the results to the client.

Tudor Dumitras - One of the best experts on this subject based on the ideXlab platform.

  • No Downtime for Data Conversions: Rethinking Hot Upgrades (CMU-PDL-09-106)
    2018
    Co-Authors: Tudor Dumitras, Priya Narasimhan
    Abstract:

    Unavailability in enterprise systems is usually the result of Planned events, such as upgrades, rather than failures. Major system upgrades entail complex data conversions that are difficult to perform on the fly, in the face of live workloads. Minimizing the Downtime imposed by such conversions is a time-intensive and error-prone manual process. We present Imago, a system that aims to simplify the upgrade process, and we show that it can eliminate all the causes of Planned Downtime recorded during the upgrade history of one of the ten most popular websites. Building on the lessons learned from past research on live upgrades in middleware systems, Imago trades off a need for additional storage resources for the ability to perform end-to-end, enterprise upgrades online, with minimal application-specific knowledge

  • Improving the Dependability of Distributed Systems through AIR Software Upgrades
    2018
    Co-Authors: Tudor Dumitras
    Abstract:

    Traditional fault-tolerance mechanisms concentrate almost entirely on responding to, avoiding, or tolerating unexpected faults or security violations. However, scheduled events, such as software upgrades, account for most of the system unavailability and often introduce data corruption or latent errors. Through two empirical studies, this dissertation identifies the leading causes of upgrade failure-breaking hidden dependencies-and of Planned Downtime -complex data conversions-in distributed enterprise systems. These findings represent the foundation of a new benchmark for software-upgrade dependability. This dissertation further introduces the AIR properties-Atomicity, Isolation and Runtime-testing-required for improving the dependability of distributed systems that undergo major software upgrades. The AIR properties are realized in Imago, a system designed to reduce both Planned and unPlanned Downtime by upgrading distributed systems end-to-end. Imago builds upon the idea of isolating the production system from the upgrade operations, in order to avoid breaking hidden dependencies and to decouple the data conversions from the normal system operation. Imago includes novel mechanisms, such as providing a parallel universe for the new version, performing data conversions opportunistically, intercepting the live workload at the ingress and egress points or executing an atomic switchover to the new version, which allow it to deliver the AIR properties. Imago harnesses opportunities provided by the emerging cloud-computing technologies, by trading resource overhead (needed by the parallel universe) for an improved dependability of the software upgrades. This approach separates the functional aspects of the upgrade from the mechanisms for online upgrade, enabling an upgrade-as-a-service model. This dissertation also describes techniques for assessing the impact of software upgrades, in order to reason about the implications of relaxing the AIR guarantees.

  • Middleware (Companion) - Toward upgrades-as-a-service in distributed systems
    2009
    Co-Authors: Tudor Dumitras, Priya Narasimhan
    Abstract:

    Unavailability in distributed enterprise systems is usually the result of Planned events, such as upgrades, rather than failures. Major system upgrades entail complex data conversions that are difficult to perform on the fly, in the face of live workloads. Minimizing the Downtime imposed by such conversions is a time-intensive and error-prone manual process. We propose upgrades-as-a-service, a novel approach that can eliminate all the causes of Planned Downtime recorded during the upgrade history of one of the ten most popular websites. Building on the lessons learned from past research on live upgrades in middleware systems, upgrades-as-a-service trade off a need for additional hardware resources during the upgrade for the ability to perform end-to-end upgrades online, with minimal application-specific knowledge.

  • OOPSLA Companion - Dependable, online upgrades in enterprise systems
    Proceeding of the 24th ACM SIGPLAN conference companion on Object oriented programming systems languages and applications - OOPSLA '09, 2009
    Co-Authors: Tudor Dumitras
    Abstract:

    Software upgrades are unreliable, often causing Downtime or data loss. I propose Imago, an approach for removing the leading causes of upgrade failures (broken dependencies) and of Planned Downtime (data migrations). While imposing a higher resource overhead than previous techniques, Imago is more dependable and easier to use correctly.

  • No Downtime for Data Conversions: Rethinking Hot Upgrades (CMU-PDL-09-106)
    2009
    Co-Authors: Tudor Dumitras, Priya Narasimhan
    Abstract:

    Unavailability in enterprise systems is usually the result of Planned events, such as upgrades, rather than failures. Major system upgrades entail complex data conversions that are difficult to perform on the fly, in the face of live workloads. Minimizing the Downtime imposed by such conversions is a time-intensive and error-prone manual process. We present Imago, a system that aims to simplify the upgrade process, and we show that it can eliminate all the causes of Planned Downtime recorded during the upgrade history of one of the ten most popular websites. Building on the lessons learned from past research on live upgrades in middleware systems, Imago trades off a need for additional storage resources for the ability to perform end-to-end, enterprise upgrades online, with minimal application-specific knowledge. Acknowledgements: We would like to thank Alan Downing, Jim Stamos and Byron Wang of Oracle for their feedback during the early stage of this research project.

Priya Narasimhan - One of the best experts on this subject based on the ideXlab platform.

  • No Downtime for Data Conversions: Rethinking Hot Upgrades (CMU-PDL-09-106)
    2018
    Co-Authors: Tudor Dumitras, Priya Narasimhan
    Abstract:

    Unavailability in enterprise systems is usually the result of Planned events, such as upgrades, rather than failures. Major system upgrades entail complex data conversions that are difficult to perform on the fly, in the face of live workloads. Minimizing the Downtime imposed by such conversions is a time-intensive and error-prone manual process. We present Imago, a system that aims to simplify the upgrade process, and we show that it can eliminate all the causes of Planned Downtime recorded during the upgrade history of one of the ten most popular websites. Building on the lessons learned from past research on live upgrades in middleware systems, Imago trades off a need for additional storage resources for the ability to perform end-to-end, enterprise upgrades online, with minimal application-specific knowledge

  • Middleware (Companion) - Toward upgrades-as-a-service in distributed systems
    2009
    Co-Authors: Tudor Dumitras, Priya Narasimhan
    Abstract:

    Unavailability in distributed enterprise systems is usually the result of Planned events, such as upgrades, rather than failures. Major system upgrades entail complex data conversions that are difficult to perform on the fly, in the face of live workloads. Minimizing the Downtime imposed by such conversions is a time-intensive and error-prone manual process. We propose upgrades-as-a-service, a novel approach that can eliminate all the causes of Planned Downtime recorded during the upgrade history of one of the ten most popular websites. Building on the lessons learned from past research on live upgrades in middleware systems, upgrades-as-a-service trade off a need for additional hardware resources during the upgrade for the ability to perform end-to-end upgrades online, with minimal application-specific knowledge.

  • No Downtime for Data Conversions: Rethinking Hot Upgrades (CMU-PDL-09-106)
    2009
    Co-Authors: Tudor Dumitras, Priya Narasimhan
    Abstract:

    Unavailability in enterprise systems is usually the result of Planned events, such as upgrades, rather than failures. Major system upgrades entail complex data conversions that are difficult to perform on the fly, in the face of live workloads. Minimizing the Downtime imposed by such conversions is a time-intensive and error-prone manual process. We present Imago, a system that aims to simplify the upgrade process, and we show that it can eliminate all the causes of Planned Downtime recorded during the upgrade history of one of the ten most popular websites. Building on the lessons learned from past research on live upgrades in middleware systems, Imago trades off a need for additional storage resources for the ability to perform end-to-end, enterprise upgrades online, with minimal application-specific knowledge. Acknowledgements: We would like to thank Alan Downing, Jim Stamos and Byron Wang of Oracle for their feedback during the early stage of this research project.

  • MetaMorphMagi: From Offline to Online Software Upgrades in Large-Scale IT Infrastructures
    2007
    Co-Authors: Tudor Dumitras, Jiaqi Tan, Priya Narasimhan
    Abstract:

    Software upgrades are one of the leading causes of Downtime in IT infrastructures. Long running datamigration processes require intensive up-front preparation, extended maintenance windows and close monitoring, and they impose a significant burden on the system administrators. Even worse, major upgrades sometimes fail due to complex, hidden dependencies within the system, causing unPlanned Downtime and loss of critical data. In this paper, we propose a technique for converting an offline data-migration process into a dependency-agnostic online upgrade that requires minimal administrative intervention and that eliminates the need for Planned Downtime. We illustrate our technique by walking the reader through a hypothetical, but realistic online upgrade scenario in a medium-sized IT infrastructure – namely, hot-swapping the wiki software that underlies Wikipedia with an entirely different wiki engine.