Presented at All Things Open 2023
Presented by Dave McAllister - nginx
Title: Know Your Data: The stats behind your alerts
Abstract: Quick, what's the difference between the mean, the mode and the median? Which mean do you mean? Do you need a Gaussian or a normal distribution And does your choice impact the alerts and observations you get from your observability tools?
Come get refreshed on the impact some basic choices in statistical behavior can have on what gets triggered. Learn why a median might be the choice for historical anomaly or sudden change. Jump into Gaussian distributions, data alignment challenges and the trouble with sampling. Walk out with a deeper understanding of your metrics and what they might be telling you.
Find more info about All Things Open:
On the web: https://www.allthingsopen.org/
Twitter: https://twitter.com/AllThingsOpen
LinkedIn: https://www.linkedin.com/company/all-things-open/
Instagram: https://www.instagram.com/allthingsopen/
Facebook: https://www.facebook.com/AllThingsOpen
Mastodon: https://mastodon.social/@allthingsopen
Threads: https://www.threads.net/@allthingsopen
2023 conference: https://2023.allthingsopen.org/
OSMC 2023 | Know your data: The stats behind your alerts by Dave McAllisterNETWAYS
Quick, what’s the difference between the mean, the mode and the median? Do you need a Gaussian or a normal distribution And does your choice impact the alerts and observations you get from your observability tools?
Come get refreshed on the impact some basic choices in statistical behavior can have on what gets triggered. Learn why a median might be the choice for historical anomaly or sudden change. Jump into Gaussian distributions, data alignment challenges, and the trouble with sampling. Walk out with a deeper understanding of your metrics and what they might tell you.
OARD 30: What part of 'NO' is so hard for the DNS to understand?APNIC
APNIC Chief Scientist Geoff Huston presents on 'What part of 'NO" is so hard for the DNS to understand' at OARC 30 in Bangkok, Thailand from 12 to 13 May 2019.
With the raise of NoSQL databases consistency models that are less strict than ACID transactions became popular again. After the first enthusiasm the developer community became aware that those relaxed consistency models hold some new challenges they never knew about in the ACID world. Fortunately there are some concepts around how to deal with those challenges. This presentation gives a rough introduction into the different consistency models that are available and their characteristics. Then it focusses on two techniques to deal with relaxed consistency. The first one are quorum-based reads and writes which provides a client-side strong consistency model (even if the database only implements eventual consistency). Then CRDTs (Conflict-free Replicated Data Types) are presented. CRDTs are self-stabilizing data structures which are designed for environments with extremely high availability requirements - and thus extremely weak consistency guarantees. Even though as always the majority of the information is on the voice track, I designed the slides in a way that they also provide some useful information without the voice.
Heuristic design of experiments w meta gradient searchGreg Makowski
Once you have started learning about predictive algorithms, and the basic knowledge discovery in databases process, what is the next level of detail to learn for a consulting project?
* Give examples of the many model training parameters
* Track results in a "model notebook"
* Use a model metric that combines both accuracy and generalization to rank models
* How to strategically search over the model training parameters - use a gradient descent approach
* One way to describe an arbitrarily complex predictive system is by using sensitivity analysis
Hardware fails, applications fail, our code... well, it fails too (at least mine). To prevent software failure we test. Hardware failures are inevitable, so we write code that tolerates them, then we test. From tests we gather metrics and act upon them by improving parts that perform inadequately. Measuring right things at right places in an application is as much about good engineering practices and maintaining SLAs as it is about end user experience and may differentiate successful product from a failure.
In order to act on performance metrics such as max latency and consistent response times we need to know their accurate value. The problem with such metrics is that when using popular tools we get results that are not only inaccurate but also too optimistic.
During my presentation I will simulate services that require monitoring and show how gathered metrics differ from real numbers. All this while using what currently seems to be most popular metric pipeline - Graphite together with metrics.dropwizard.io library - and get completely false results. We will learn to tune it and get much better accuracy. We will use JMeter to measure latency and observe how falsely reassuring the results are. Finally I will show how HdrHistogram helps in gathering reliable metrics. We will also run tests measuring performance of different metric classes.
This document discusses how to conduct spot speed studies to collect traffic speed data. It outlines the objectives of spot speed studies which include determining characteristics like the mean, median, mode and 85th percentile speed. It describes different speed study considerations and parameters of interest. It also covers how to analyze spot speed study data, check if the speed distribution is normal, and how to determine the required sample size.
As your service footprint grows, adding traffic control capabilities beyond stock solutions like kube-proxy becomes critical. Envoy provides fine grained routing control, load shedding, and metrics that help you scale your environment smoothly. We'll walk through several traffic control strategies using Envoy.
Presented at All Things Open 2023
Presented by Dave McAllister - nginx
Title: Know Your Data: The stats behind your alerts
Abstract: Quick, what's the difference between the mean, the mode and the median? Which mean do you mean? Do you need a Gaussian or a normal distribution And does your choice impact the alerts and observations you get from your observability tools?
Come get refreshed on the impact some basic choices in statistical behavior can have on what gets triggered. Learn why a median might be the choice for historical anomaly or sudden change. Jump into Gaussian distributions, data alignment challenges and the trouble with sampling. Walk out with a deeper understanding of your metrics and what they might be telling you.
Find more info about All Things Open:
On the web: https://www.allthingsopen.org/
Twitter: https://twitter.com/AllThingsOpen
LinkedIn: https://www.linkedin.com/company/all-things-open/
Instagram: https://www.instagram.com/allthingsopen/
Facebook: https://www.facebook.com/AllThingsOpen
Mastodon: https://mastodon.social/@allthingsopen
Threads: https://www.threads.net/@allthingsopen
2023 conference: https://2023.allthingsopen.org/
OSMC 2023 | Know your data: The stats behind your alerts by Dave McAllisterNETWAYS
Quick, what’s the difference between the mean, the mode and the median? Do you need a Gaussian or a normal distribution And does your choice impact the alerts and observations you get from your observability tools?
Come get refreshed on the impact some basic choices in statistical behavior can have on what gets triggered. Learn why a median might be the choice for historical anomaly or sudden change. Jump into Gaussian distributions, data alignment challenges, and the trouble with sampling. Walk out with a deeper understanding of your metrics and what they might tell you.
OARD 30: What part of 'NO' is so hard for the DNS to understand?APNIC
APNIC Chief Scientist Geoff Huston presents on 'What part of 'NO" is so hard for the DNS to understand' at OARC 30 in Bangkok, Thailand from 12 to 13 May 2019.
With the raise of NoSQL databases consistency models that are less strict than ACID transactions became popular again. After the first enthusiasm the developer community became aware that those relaxed consistency models hold some new challenges they never knew about in the ACID world. Fortunately there are some concepts around how to deal with those challenges. This presentation gives a rough introduction into the different consistency models that are available and their characteristics. Then it focusses on two techniques to deal with relaxed consistency. The first one are quorum-based reads and writes which provides a client-side strong consistency model (even if the database only implements eventual consistency). Then CRDTs (Conflict-free Replicated Data Types) are presented. CRDTs are self-stabilizing data structures which are designed for environments with extremely high availability requirements - and thus extremely weak consistency guarantees. Even though as always the majority of the information is on the voice track, I designed the slides in a way that they also provide some useful information without the voice.
Heuristic design of experiments w meta gradient searchGreg Makowski
Once you have started learning about predictive algorithms, and the basic knowledge discovery in databases process, what is the next level of detail to learn for a consulting project?
* Give examples of the many model training parameters
* Track results in a "model notebook"
* Use a model metric that combines both accuracy and generalization to rank models
* How to strategically search over the model training parameters - use a gradient descent approach
* One way to describe an arbitrarily complex predictive system is by using sensitivity analysis
Hardware fails, applications fail, our code... well, it fails too (at least mine). To prevent software failure we test. Hardware failures are inevitable, so we write code that tolerates them, then we test. From tests we gather metrics and act upon them by improving parts that perform inadequately. Measuring right things at right places in an application is as much about good engineering practices and maintaining SLAs as it is about end user experience and may differentiate successful product from a failure.
In order to act on performance metrics such as max latency and consistent response times we need to know their accurate value. The problem with such metrics is that when using popular tools we get results that are not only inaccurate but also too optimistic.
During my presentation I will simulate services that require monitoring and show how gathered metrics differ from real numbers. All this while using what currently seems to be most popular metric pipeline - Graphite together with metrics.dropwizard.io library - and get completely false results. We will learn to tune it and get much better accuracy. We will use JMeter to measure latency and observe how falsely reassuring the results are. Finally I will show how HdrHistogram helps in gathering reliable metrics. We will also run tests measuring performance of different metric classes.
This document discusses how to conduct spot speed studies to collect traffic speed data. It outlines the objectives of spot speed studies which include determining characteristics like the mean, median, mode and 85th percentile speed. It describes different speed study considerations and parameters of interest. It also covers how to analyze spot speed study data, check if the speed distribution is normal, and how to determine the required sample size.
As your service footprint grows, adding traffic control capabilities beyond stock solutions like kube-proxy becomes critical. Envoy provides fine grained routing control, load shedding, and metrics that help you scale your environment smoothly. We'll walk through several traffic control strategies using Envoy.
Initially presented at Software Architecture Conference in Boston, MA on 3/18/15.
Distributed systems are complex beasts. Breaking your application into multiple services introduces new types of errors, cascading failures, and CAP theorem limitations. Unfortunately, your uptime and sanity both suffer. This session will focus on various tactics and learnings from Lucid Software's migration to a service oriented architecture.
Finding Bugs Faster with Assertion Based Verification (ABV)DVClub
1) Assertion-based verification introduces assertions into a design to improve observability and controllability during simulation and formal analysis.
2) Assertions define expected behavior and can detect errors by monitoring signals within a design.
3) An assertion-based verification methodology leverages assertions throughout the verification flow from module to system level using various tools like simulation, formal analysis, and acceleration for improved productivity, quality, and reduced verification time.
Reaching Consensus in Crowdsourced Transcription of Biocollections Information
Andréa Matsunaga (ammatsun@ufl.edu), Austin Mast, and José A.B. Fortes
10th IEEE International Conference on e-Science
October 23, 2014
Guarujá, SP, Brazil
EF-1115210
This document evaluates the Nebraska Department of Roads' (NDOR) Actuated Advance Warning (AAW) system at signalized intersections. It analyzes safety and operational data from 26 intersections with AAW systems and 29 comparison intersections. Statistical models show the AAW system likely reduces crashes by over 90%. Operational analyses found the system reduces the number of vehicles trapped in the "dilemma zone" and decreases the frequency that lights reach maximum time. Microsimulation models were developed and validated for two test sites. A sensitivity analysis examined how factors like turn percentage affect average wait times and conflicts. The conclusions recommend the AAW system for other high-speed intersections and provide guidelines for when to install or remove the systems based on measures like
Performance Issue? Machine Learning to the rescue!Maarten Smeets
t can be difficult to determine how to improve performance of microservices. There are many factors you can vary but which factor will be the one having most impact? During this presentation, a method using the random forest machine learning algorithm will be applied in order to help improve performance of a microservice running inside a JVM. Several measures are taken such as thoughput and response times. Java version, JVM supplier, heap, garbage collection algorithm and microservice framework are all varied. Which factor is most important in determining the response time and throughput of the services? The Random Forest algorithm will be introduced to solve this challenge. Not only will this presentation give some useful suggestions for improving the performance of microservices but will also introduce a novel way to take on the challenge of performance tuning which can be applied to other use-cases. This presentation is especially interesting to developers and architects.
This document proposes fast single-pass k-means clustering algorithms to allow for fast nearest neighbor search on large datasets. It discusses the rationale for using k-means clustering, describes algorithms like ball k-means and surrogate methods that can perform clustering in a single pass. It covers implementations using techniques like locality sensitive hashing and projection search to speed up vector searches. Evaluation on synthetic and real datasets shows the algorithms can achieve the same or better accuracy as traditional k-means 10x faster, enabling applications like fast nearest neighbor search on massive datasets for applications like customer modeling.
Sean Kandel - Data profiling: Assessing the overall content and quality of a ...huguk
The task of “data profiling”—assessing the overall content and quality of a data set—is a core aspect of the analytic experience. Traditionally, profiling was a fairly cut-and-dried task: load the raw numbers into a stat package, run some basic descriptive statistics, and report the output in a summary file or perhaps a simple data visualization. However, data volumes can be so large today that traditional tools and methods for computing descriptive statistics become intractable; even with scalable infrastructure like Hadoop, aggressive optimization and statistical approximation techniques must be used. In this talk Sean will cover technical challenges in keeping data profiling agile in the Big Data era. He will discuss both research results and real-world best practices used by analysts in the field, including methods for sampling, summarizing and sketching data, and the pros and cons of using these various approaches.
Sean is Trifacta’s Chief Technical Officer. He completed his Ph.D. at Stanford University, where his research focused on user interfaces for database systems. At Stanford, Sean led development of new tools for data transformation and discovery, such as Data Wrangler. He previously worked as a data analyst at Citadel Investment Group.
Bridging the Gap: Machine Learning for Ubiquitous Computing -- EvaluationThomas Ploetz
Tutorial @Ubicomp 2015: Bridging the Gap -- Machine Learning for Ubiquitous Computing (evaluation session).
A tutorial on promises and pitfalls of Machine Learning for Ubicomp (and Human Computer Interaction). From Practitioners for Practitioners.
Presenter: Nils Hammerla <n.hammerla@gmail.com>
video recording of talks as they wer held at Ubicomp:
https://youtu.be/LgnnlqOIXJc?list=PLh96aGaacSgXw0MyktFqmgijLHN-aQvdq
The document discusses research on learning to improve the efficiency of machine learning algorithms through speedup learning. It provides three key points:
1) Early work on explanation-based learning for speedup had limited success, but techniques like memoization and clause learning led to major improvements in SAT solvers.
2) More recent approaches use machine learning to build predictive models of problem instances and solver behavior, in order to inform strategies like automatic noise setting and randomized restart policies.
3) Case studies demonstrate these learning-based approaches can outperform traditional techniques and fixed policies by customizing resource allocation and reformulation based on problem structure and solver progress.
The document discusses research on learning to improve the efficiency of machine learning algorithms through speedup learning. It provides three key points:
1) Early work on explanation-based learning for speedup had limited success, but techniques like memoization and clause learning led to major improvements in SAT solvers.
2) More recent approaches use predictive models trained on dynamic features to learn optimal policies for controlling search algorithms, like setting noise levels or restart policies.
3) Open problems remain in developing optimal predictive policies with partial information and approximations, to continue improving search and reasoning performance.
(BDT207) Real-Time Analytics In Service Of Self-Healing EcosystemsAmazon Web Services
Netflix strives to provide an amazing experience to each member. To accomplish this, Netflix needs to maintain very high availability across our systems. However, at a certain scale, humans can no longer scale their ability to monitor the status of all systems, making it critical for Netflix to build tools and platforms that can automatically monitor their production environments and make intelligent real-time operational decisions to remedy the problems they identify. In this session, we discuss how Netflix uses data mining and machine learning techniques to automate decisions in real-time with the goal of supporting operational availability, reliability, and consistency. We review how we got to the current states, the lessons we learned, and the future of real-time analytics at Netflix. While Netflix's scale is larger than most other companies, we believe the approaches and technologies we discuss are highly relevant to other production environments, and audience members should come away with actionable ideas that are implementable in, and benefit, most other environments.
Decision Forests and discriminant analysispotaters
This document summarizes a tutorial on randomised decision forests and tree-structured algorithms. It discusses how tree-based algorithms like boosting and random forests can be used for tasks like object detection, tracking and segmentation. It also describes techniques for speeding up computation, such as converting boosted classifiers to decision trees and using multiple classifier systems. The tutorial is structured in two parts, covering tree-structured algorithms and randomised forests.
This document discusses various topics related to optimizing OLTP performance in Oracle databases, including:
1) Database performance principles such as acceptable CPU utilization levels and how user response times are affected by utilization levels above 60-65%.
2) Different connection architectures including dedicated servers, shared servers, and database resident connection pooling and their tradeoffs in terms of connection speed and code path length.
3) The importance of writing efficient SQL statements and maintaining proper schema statistics to enable the database to choose efficient execution plans.
4) Best practices for SQL optimization such as validating join conditions, indexes, partition pruning strategies, and parallelization levels are emphasized.
The document describes several parallel approaches for speeding up sequence alignment. It discusses splitting a DNA string into chunks that are distributed to worker nodes. Various techniques are proposed for handling matches that span multiple chunks, including using bigger chunks, on-demand requests for additional data, and having workers find partial matches along chunk edges to be combined by the master. The approaches are analyzed in terms of advantages and disadvantages, and a test plan is outlined to evaluate performance under varying parameters like number of workers and query length.
This document discusses factors that influence web search latency from both the user and system perspectives. It summarizes that users expect fast response times from search engines, while search engines aim to balance speed, quality, and costs. The document then outlines components that contribute to latency, experiments measuring user sensitivity to latency, and the impact of latency on user search experience. Specifically, it finds users notice delays over 1000ms and that faster search sites lead to higher user engagement.
Beyond Averages - Web Performance MeetupDan Kuebrich
When raw data becomes overwhelming, we turn to abstraction to understand our world. In examining the performance of our systems, the data is always overwhelming. Solutions like summary statistics have come to our rescue, and they are good—up to a point. In order to truly understand our systems, we need to know when and how to sidestep those abstractions,to get deep, detailed performance insight. At this meetup, I’ll explore techniques for visualizing the underlying structure of performance data and how this empowers drilling down to populations and individual samples in the data set.
Towards Detecting Performance Anti-patterns Using Classification TechniquesJames Hill
This is the talk I gave on behalf of my Ph.D. student at the Machine Learning and Information Retrieval (MALIR) for Software Evolution (MALIR-SE) workshop at ASE 2013.
Splunk Enterprise for Information Security Hands-On Breakout SessionSplunk
The document provides information about detecting various types of cyber attacks using Splunk, including web attacks, lateral movement, and DNS exfiltration. It includes examples of search queries and apps that can be used in Splunk to detect SQL injection, pass-the-hash attacks, and DNS tunneling used for data exfiltration. The document demonstrates how machine data from different sources can be analyzed in Splunk to gain visibility into attack behaviors and detect security incidents.
Initially presented at Software Architecture Conference in Boston, MA on 3/18/15.
Distributed systems are complex beasts. Breaking your application into multiple services introduces new types of errors, cascading failures, and CAP theorem limitations. Unfortunately, your uptime and sanity both suffer. This session will focus on various tactics and learnings from Lucid Software's migration to a service oriented architecture.
Finding Bugs Faster with Assertion Based Verification (ABV)DVClub
1) Assertion-based verification introduces assertions into a design to improve observability and controllability during simulation and formal analysis.
2) Assertions define expected behavior and can detect errors by monitoring signals within a design.
3) An assertion-based verification methodology leverages assertions throughout the verification flow from module to system level using various tools like simulation, formal analysis, and acceleration for improved productivity, quality, and reduced verification time.
Reaching Consensus in Crowdsourced Transcription of Biocollections Information
Andréa Matsunaga (ammatsun@ufl.edu), Austin Mast, and José A.B. Fortes
10th IEEE International Conference on e-Science
October 23, 2014
Guarujá, SP, Brazil
EF-1115210
This document evaluates the Nebraska Department of Roads' (NDOR) Actuated Advance Warning (AAW) system at signalized intersections. It analyzes safety and operational data from 26 intersections with AAW systems and 29 comparison intersections. Statistical models show the AAW system likely reduces crashes by over 90%. Operational analyses found the system reduces the number of vehicles trapped in the "dilemma zone" and decreases the frequency that lights reach maximum time. Microsimulation models were developed and validated for two test sites. A sensitivity analysis examined how factors like turn percentage affect average wait times and conflicts. The conclusions recommend the AAW system for other high-speed intersections and provide guidelines for when to install or remove the systems based on measures like
Performance Issue? Machine Learning to the rescue!Maarten Smeets
t can be difficult to determine how to improve performance of microservices. There are many factors you can vary but which factor will be the one having most impact? During this presentation, a method using the random forest machine learning algorithm will be applied in order to help improve performance of a microservice running inside a JVM. Several measures are taken such as thoughput and response times. Java version, JVM supplier, heap, garbage collection algorithm and microservice framework are all varied. Which factor is most important in determining the response time and throughput of the services? The Random Forest algorithm will be introduced to solve this challenge. Not only will this presentation give some useful suggestions for improving the performance of microservices but will also introduce a novel way to take on the challenge of performance tuning which can be applied to other use-cases. This presentation is especially interesting to developers and architects.
This document proposes fast single-pass k-means clustering algorithms to allow for fast nearest neighbor search on large datasets. It discusses the rationale for using k-means clustering, describes algorithms like ball k-means and surrogate methods that can perform clustering in a single pass. It covers implementations using techniques like locality sensitive hashing and projection search to speed up vector searches. Evaluation on synthetic and real datasets shows the algorithms can achieve the same or better accuracy as traditional k-means 10x faster, enabling applications like fast nearest neighbor search on massive datasets for applications like customer modeling.
Sean Kandel - Data profiling: Assessing the overall content and quality of a ...huguk
The task of “data profiling”—assessing the overall content and quality of a data set—is a core aspect of the analytic experience. Traditionally, profiling was a fairly cut-and-dried task: load the raw numbers into a stat package, run some basic descriptive statistics, and report the output in a summary file or perhaps a simple data visualization. However, data volumes can be so large today that traditional tools and methods for computing descriptive statistics become intractable; even with scalable infrastructure like Hadoop, aggressive optimization and statistical approximation techniques must be used. In this talk Sean will cover technical challenges in keeping data profiling agile in the Big Data era. He will discuss both research results and real-world best practices used by analysts in the field, including methods for sampling, summarizing and sketching data, and the pros and cons of using these various approaches.
Sean is Trifacta’s Chief Technical Officer. He completed his Ph.D. at Stanford University, where his research focused on user interfaces for database systems. At Stanford, Sean led development of new tools for data transformation and discovery, such as Data Wrangler. He previously worked as a data analyst at Citadel Investment Group.
Bridging the Gap: Machine Learning for Ubiquitous Computing -- EvaluationThomas Ploetz
Tutorial @Ubicomp 2015: Bridging the Gap -- Machine Learning for Ubiquitous Computing (evaluation session).
A tutorial on promises and pitfalls of Machine Learning for Ubicomp (and Human Computer Interaction). From Practitioners for Practitioners.
Presenter: Nils Hammerla <n.hammerla@gmail.com>
video recording of talks as they wer held at Ubicomp:
https://youtu.be/LgnnlqOIXJc?list=PLh96aGaacSgXw0MyktFqmgijLHN-aQvdq
The document discusses research on learning to improve the efficiency of machine learning algorithms through speedup learning. It provides three key points:
1) Early work on explanation-based learning for speedup had limited success, but techniques like memoization and clause learning led to major improvements in SAT solvers.
2) More recent approaches use machine learning to build predictive models of problem instances and solver behavior, in order to inform strategies like automatic noise setting and randomized restart policies.
3) Case studies demonstrate these learning-based approaches can outperform traditional techniques and fixed policies by customizing resource allocation and reformulation based on problem structure and solver progress.
The document discusses research on learning to improve the efficiency of machine learning algorithms through speedup learning. It provides three key points:
1) Early work on explanation-based learning for speedup had limited success, but techniques like memoization and clause learning led to major improvements in SAT solvers.
2) More recent approaches use predictive models trained on dynamic features to learn optimal policies for controlling search algorithms, like setting noise levels or restart policies.
3) Open problems remain in developing optimal predictive policies with partial information and approximations, to continue improving search and reasoning performance.
(BDT207) Real-Time Analytics In Service Of Self-Healing EcosystemsAmazon Web Services
Netflix strives to provide an amazing experience to each member. To accomplish this, Netflix needs to maintain very high availability across our systems. However, at a certain scale, humans can no longer scale their ability to monitor the status of all systems, making it critical for Netflix to build tools and platforms that can automatically monitor their production environments and make intelligent real-time operational decisions to remedy the problems they identify. In this session, we discuss how Netflix uses data mining and machine learning techniques to automate decisions in real-time with the goal of supporting operational availability, reliability, and consistency. We review how we got to the current states, the lessons we learned, and the future of real-time analytics at Netflix. While Netflix's scale is larger than most other companies, we believe the approaches and technologies we discuss are highly relevant to other production environments, and audience members should come away with actionable ideas that are implementable in, and benefit, most other environments.
Decision Forests and discriminant analysispotaters
This document summarizes a tutorial on randomised decision forests and tree-structured algorithms. It discusses how tree-based algorithms like boosting and random forests can be used for tasks like object detection, tracking and segmentation. It also describes techniques for speeding up computation, such as converting boosted classifiers to decision trees and using multiple classifier systems. The tutorial is structured in two parts, covering tree-structured algorithms and randomised forests.
This document discusses various topics related to optimizing OLTP performance in Oracle databases, including:
1) Database performance principles such as acceptable CPU utilization levels and how user response times are affected by utilization levels above 60-65%.
2) Different connection architectures including dedicated servers, shared servers, and database resident connection pooling and their tradeoffs in terms of connection speed and code path length.
3) The importance of writing efficient SQL statements and maintaining proper schema statistics to enable the database to choose efficient execution plans.
4) Best practices for SQL optimization such as validating join conditions, indexes, partition pruning strategies, and parallelization levels are emphasized.
The document describes several parallel approaches for speeding up sequence alignment. It discusses splitting a DNA string into chunks that are distributed to worker nodes. Various techniques are proposed for handling matches that span multiple chunks, including using bigger chunks, on-demand requests for additional data, and having workers find partial matches along chunk edges to be combined by the master. The approaches are analyzed in terms of advantages and disadvantages, and a test plan is outlined to evaluate performance under varying parameters like number of workers and query length.
This document discusses factors that influence web search latency from both the user and system perspectives. It summarizes that users expect fast response times from search engines, while search engines aim to balance speed, quality, and costs. The document then outlines components that contribute to latency, experiments measuring user sensitivity to latency, and the impact of latency on user search experience. Specifically, it finds users notice delays over 1000ms and that faster search sites lead to higher user engagement.
Beyond Averages - Web Performance MeetupDan Kuebrich
When raw data becomes overwhelming, we turn to abstraction to understand our world. In examining the performance of our systems, the data is always overwhelming. Solutions like summary statistics have come to our rescue, and they are good—up to a point. In order to truly understand our systems, we need to know when and how to sidestep those abstractions,to get deep, detailed performance insight. At this meetup, I’ll explore techniques for visualizing the underlying structure of performance data and how this empowers drilling down to populations and individual samples in the data set.
Towards Detecting Performance Anti-patterns Using Classification TechniquesJames Hill
This is the talk I gave on behalf of my Ph.D. student at the Machine Learning and Information Retrieval (MALIR) for Software Evolution (MALIR-SE) workshop at ASE 2013.
Splunk Enterprise for Information Security Hands-On Breakout SessionSplunk
The document provides information about detecting various types of cyber attacks using Splunk, including web attacks, lateral movement, and DNS exfiltration. It includes examples of search queries and apps that can be used in Splunk to detect SQL injection, pass-the-hash attacks, and DNS tunneling used for data exfiltration. The document demonstrates how machine data from different sources can be analyzed in Splunk to gain visibility into attack behaviors and detect security incidents.
Similar to Statistical Analysis of DNS Latencies.pdf (20)
Gen Z and the marketplaces - let's translate their needsLaura Szabó
The product workshop focused on exploring the requirements of Generation Z in relation to marketplace dynamics. We delved into their specific needs, examined the specifics in their shopping preferences, and analyzed their preferred methods for accessing information and making purchases within a marketplace. Through the study of real-life cases , we tried to gain valuable insights into enhancing the marketplace experience for Generation Z.
The workshop was held on the DMA Conference in Vienna June 2024.
Ready to Unlock the Power of Blockchain!Toptal Tech
Imagine a world where data flows freely, yet remains secure. A world where trust is built into the fabric of every transaction. This is the promise of blockchain, a revolutionary technology poised to reshape our digital landscape.
Toptal Tech is at the forefront of this innovation, connecting you with the brightest minds in blockchain development. Together, we can unlock the potential of this transformative technology, building a future of transparency, security, and endless possibilities.
3. DNS Performance Metrics (quick intro) Measuring DNS Latency
• Performance under normal conditions
‣
• The data is right-skewed
‣ The usual descriptive statistics are useless (average, mean, …)
‣ Most of the queries are answered very quickly
‣ In fact, 95% of the queries are answered under 2 milliseconds
‣ The tails makes it interesting
Ondřej Surý <ondrej@isc.org> 2024-05-21 1 / 8
5. Logarithmic Percentile Histogram Measuring DNS Latency
• Both axes are logarithmic
‣ x-axis: slowest percentile
‣ y-axis: average latency
• It makes the tail more visible
• Variant of Complementary Cumulative Distribution Function
• Very robust, can be used for monitoring (1% slowest percentile)
• Introduced by the good folks at PowerDNS
See more: https://blog.powerdns.com/2017/11/02/dns-performance-
metrics-the-logarithmic-percentile-histogram
Ondřej Surý <ondrej@isc.org> 2024-05-21 3 / 8
7. DNS Performance for Developers Compare DNS Latencies
• Comparing two branches of BIND 9
‣ Did we improve the code?
‣ Did we made things worse?
‣ Currently, we compare the graphs by looking at them;
‣ And then running more tests;
‣ And then some wishful thinking…
• Sending thanks to Python’s numpy and scipy developers!
Ondřej Surý <ondrej@isc.org> 2024-05-21 4 / 8
8. Pick the right statistics Compare DNS Latencies
• The distribution is not normal
• Non-parametrical test then?
‣ Kolmogorov-Smirnov test didn’t really work
• Normalize the data?
‣ Box Cox Transformation didn’t really work
• Maybe look only at the tail then?
Ondřej Surý <ondrej@isc.org> 2024-05-21 5 / 8
9. Looking at the tail Compare DNS Latencies
• Pick the 95% (99%) percentile complement
‣ Either return the lowest bucket needed for 5% of responses
‣ Or count the answers in (1.9-2.0 second buckets)
• Have at least 3 runs for each group
• Yay! The data are normal and the group variances are equal
‣ Shapiro-Wilk test
– first group (𝑊 = 0.905, 𝑝 = 0.436)
– second group (𝑊 = 0.970, 𝑝 = 0.874)
‣ Brown-Forsyth test (𝐹 = 0.070, 𝑝 = 0.798)
Ondřej Surý <ondrej@isc.org> 2024-05-21 6 / 8
10. Parametrical test (ANOVA) Compare DNS Latencies
• We can test more than two branches
• One-way ANOVA reports difference between branches
‣ 𝐹 = 9244.090, 𝑝 < .001
• Two-sample T-Test (for confirmation)
‣ 𝑇 = −96.146, 𝑝 < .001
Ondřej Surý <ondrej@isc.org> 2024-05-21 7 / 8
12. Other tests? More ideas?
• Is this even correct? Or am I crazy? (I’m not a statistician)
• Can we just compare two data sets (1x baseline with 1x branch)?
• Can we use the full (right-skewed) population?
• Are there any other non-parametrical tests I can try/use?
• Are there any other suitable statistical methods?
• Is this useful for other Internet measurements?
Ondřej Surý <ondrej@isc.org> 2024-05-21 8 / 8