Disk Array Performance Estimation AGH University of Science and Technology Department of Computer Science Jacek Marmuszewski Darin Nikołow, Marek Pogoda,

Slides:



Advertisements
Similar presentations
IBM Software Group ® Integrated Server and Virtual Storage Management an IT Optimization Infrastructure Solution from IBM Small and Medium Business Software.
Advertisements

IS 4506 Tuning and Monitoring Internet Information Server.
By the end of this section, you will know and understand the hardware and software involved in making a LAN!
A New Cache Management Approach for Transaction Processing on Flash-based Database Da Zhou
1 GridTorrent Framework: A High-performance Data Transfer and Data Sharing Framework for Scientific Computing.
Esma Yildirim Department of Computer Engineering Fatih University Istanbul, Turkey DATACLOUD 2013.
Enhanced Availability With RAID CC5493/7493. RAID Redundant Array of Independent Disks RAID is implemented to improve: –IO throughput (speed) and –Availability.
Linux Clustering A way to supercomputing. What is Cluster? A group of individual computers bundled together using hardware and software in order to make.
Optimizing of data access using replication technique Renata Słota 1, Darin Nikolow 1,Łukasz Skitał 2, Jacek Kitowski 1,2 1 Institute of Computer Science.
1 Virtual Machine Resource Monitoring and Networking of Virtual Machines Ananth I. Sundararaj Department of Computer Science Northwestern University July.
1 Searching the Web Junghoo Cho UCLA Computer Science.
SUMS Storage Requirement 250 TB fixed disk cache 130 TB annual increment for permanently on- line data 100 TB work area (not controlled by SUMS) 2 PB near-line.
Implementing ISA Server Caching. Caching Overview ISA Server supports caching as a way to improve the speed of retrieving information from the Internet.
© 2009 IBM Corporation Statements of IBM future plans and directions are provided for information purposes only. Plans and direction are subject to change.
Virtual Network Servers. What is a Server? 1. A software application that provides a specific one or more services to other computers  Example: Apache.
Amin Kazempour Long Yunyan XU
Gordon: Using Flash Memory to Build Fast, Power-efficient Clusters for Data-intensive Applications A. Caulfield, L. Grupp, S. Swanson, UCSD, ASPLOS’09.
Capacity Planning in SharePoint Capacity Planning Process of evaluating a technology … Deciding … Hardware … Variety of Ways Different Services.
Chapter 7 Configuring & Managing Distributed File System
1. Outline Introduction Virtualization Platform - Hypervisor High-level NAS Functions Applications Supported NAS models 2.
Cluster computing facility for CMS simulation work at NPD-BARC Raman Sehgal.
Nelson Androes Online Achievement Level Setting Software.
A Workflow-Aware Storage System Emalayan Vairavanathan 1 Samer Al-Kiswany, Lauro Beltrão Costa, Zhao Zhang, Daniel S. Katz, Michael Wilde, Matei Ripeanu.
Virtual Organization Approach for Running HEP Applications in Grid Environment Łukasz Skitał 1, Łukasz Dutka 1, Renata Słota 2, Krzysztof Korcyl 3, Maciej.
Polish Infrastructure for Supporting Computational Science in the European Research Space Policy Driven Data Management in PL-Grid Virtual Organizations.
Report : Zhen Ming Wu 2008 IEEE 9th Grid Computing Conference.
PayDox Corporate Document Management System Rotech AB Interface Ltd Business Software Integration.
Step Arena Storage Introduction. 2 HDD trend- SAS is the future Source: (IDC) Infostor June 2008.
Polish Infrastructure for Supporting Computational Science in the European Research Space QoS provisioning for data-oriented applications in PL-Grid D.
Łukasz Skitał 2, Renata Słota 1, Maciej Janusz 1 and Jacek Kitowski 1,2 1 Institute of Computer Science AGH University of Science and Technology, Mickiewicza.
A Web Crawler Design for Data Mining
Principles of Scalable HPC System Design March 6, 2012 Sue Kelly Sandia National Laboratories Abstract: Sandia National.
EXPOSE GOOGLE APP ENGINE AS TASKTRACKER NODES AND DATA NODES.
Indiana University’s Name for its Sakai Implementation Oncourse CL (Collaborative Learning) Active Users = 112,341 Sites.
The Red Storm High Performance Computer March 19, 2008 Sue Kelly Sandia National Laboratories Abstract: Sandia National.
Politecnico di Torino Dipartimento di Automatica ed Informatica TORSEC Group Performance of Xen’s Secured Virtual Networks Emanuele Cesena Paolo Carlo.
1 Selecting LAN server (Week 3, Monday 9/8/2003) © Abdou Illia, Fall 2003.
High Performance Computing Processors Felix Noble Mirayma V. Rodriguez Agnes Velez Electric and Computer Engineer Department August 25, 2004.
Lecture 16: Storage and I/O EEN 312: Processors: Hardware, Software, and Interfacing Department of Electrical and Computer Engineering Spring 2014, Dr.
Introduction to dCache Zhenping (Jane) Liu ATLAS Computing Facility, Physics Department Brookhaven National Lab 09/12 – 09/13, 2005 USATLAS Tier-1 & Tier-2.
Switched Storage Architecture Benefits Computer Measurements Group November 14 th, 2002 Yves Coderre.
I/O Computer Organization II 1 Introduction I/O devices can be characterized by – Behavior: input, output, storage – Partner: human or machine – Data rate:
Price Performance Metrics CS3353. CPU Price Performance Ratio Given – Average of 6 clock cycles per instruction – Clock rating for the cpu – Number of.
1 CS : Technology Trends Ion Stoica and Ali Ghodsi ( August 31, 2015.
KUKDM’2011, Zakopane Semantic Based Storage QoS Management Methodology Renata Słota, Darin Nikolow, Jacek Kitowski Institute of Computer Science AGH-UST,
 The End to the Means › (According to IBM ) › 03.ibm.com/innovation/us/thesmartercity/in dex_flash.html?cmp=blank&cm=v&csr=chap ter_edu&cr=youtube&ct=usbrv111&cn=agus.
Rafał Słota, Michał Wrzeszcz, Renata G. Słota, Łukasz Dutka, Jacek Kitowski ACC Cyfronet AGH Department of Computer Science, AGH - UST CGW 2015 Kraków,
Communications & Networks National 4 & 5 Computing Science.
Lecture 5: Memory Performance. Types of Memory Registers L1 cache L2 cache L3 cache Main Memory Local Secondary Storage (local disks) Remote Secondary.
ICC Module 3 Lesson 3 – Storage 1 / 4 © 2015 Ph. Janson Information, Computing & Communication Storage – Clip 0 – Introduction School of Computer Science.
Metadata Organization and Management for Globalization of Data Access with Michał Wrzeszcz, Krzysztof Trzepla, Rafał Słota, Konrad Zemek, Tomasz Lichoń,
1 CEG 2400 Fall 2012 Network Servers. 2 Network Servers Critical Network servers – Contain redundant components Power supplies Fans Memory CPU Hard Drives.
AMS02 Software and Hardware Evaluation A.Eline. Outline  AMS SOC  AMS POC  AMS Gateway Computer  AMS Servers  AMS ProductionNodes  AMS Backup Solution.
HP Proliant Server  Intel Xeon E3-1220v3 (3.1GHz / 4-core / 8MB / 80W).  HP 4GB Dual Rank x8 PC E (DDR3-1600) Unbuffered Memory Kit.  HP Ethernet.
Simulation Production System Science Advisory Committee Meeting UW-Madison March 1 st -2 nd 2007 Juan Carlos Díaz Vélez.
LIOProf: Exposing Lustre File System Behavior for I/O Middleware
Storage System Optimization. Introduction Storage Types-DAS/NAS/SAN The purposes of different RAID types. How to calculate the storage size for video.
Enhanced Availability With RAID CC5493/7493. RAID Redundant Array of Independent Disks RAID is implemented to improve: –IO throughput (speed) and –Availability.
Dynamic and Scalable Distributed Metadata Management in Gluster File System Huang Qiulan Computing Center,Institute of High Energy Physics,
Configuring SQL Server for a successful SharePoint Server Deployment Haaron Gonzalez Solution Architect & Consultant Microsoft MVP SharePoint Server
AS 61/62 USP The 1 st BRASWELL NAS (Celeron N3050/N3150) – Best performance: 224 MB/s read, 213 MB/s write Equipped with dedicated hardware encryption.
G. Russo, D. Del Prete, S. Pardi Frascati, 2011 april 4th-7th The Naples' testbed for the SuperB computing model: first tests G. Russo, D. Del Prete, S.
A Web Based Job Submission System for a Physics Computing Cluster David Jones IOP Particle Physics 2004 Birmingham 1.
SOURCE:2014 IEEE 17TH INTERNATIONAL CONFERENCE ON COMPUTATIONAL SCIENCE AND ENGINEERING AUTHER: MINGLIU LIU, DESHI LI, HAILI MAO SPEAKER: JIAN-MING HONG.
The demonstration of Lustre in EAST data system
Control and data acquisition system of the KTX device
VI-SEEM Data Discovery Service
Software Architecture in Practice
Design Unit 26 Design a small or home office network
Introduction to Operating Systems
Presentation transcript:

Disk Array Performance Estimation AGH University of Science and Technology Department of Computer Science Jacek Marmuszewski Darin Nikołow, Marek Pogoda, Renata Slota, Jacek Kitowski

Outline Introduction  motivation for performance estimation of disk array  problems connected with estimation of disk array  requirements for disk array estimator Solution Environment Disk Array Performance Tests Estimation Model Estimation Quality Tests Future work

Introduction motivation for performance estimation of disk array Proper and efficient performance estimation of storage systems is essential for many processes occurring in distributed computational environments such as:  replica selection,  new replica creation,  creating VO specifying data storage performance requirements in SLA,  guarantying the fulfillment of SLA within VO.

Introduction problems connected with estimation of disk array Complexity of algorithms used to determine best solution for storing data Shared resources Virtualization

Introduction requirements for disk array estimator  Estimator response time  Estimation quality

Solution Model identification via active experiments

Environment general view - Disk Array - Host / Server - User / Application

Test Environment Disk Array 1 Infortrend A16F-G2430 2GB cache 16x 1TB HDD – SATA, RAID6 2x 4/8 Gbit/s fiber channel interface Server 1 Xeon QuadCore 4GB RAM Disk Array 2 Intel Entery Storage System SS4200-E 4 x 500GB HDD - SATA RAID5 1 Gbit Ethernet Server 2 Intel Core2Duo E Ghz 2GB RAM

Test Environment - Disk Array - Host / Server Monitoring daemon Estimator Service & database Sending data using ICE ICE – Internet Communication Engine form ZeroC

Disk Array Performance Tests Tests written in C Using 'fwrite' Synchronizing (flushing) once – before ending test usu

Disk Array Performance Tests

Model How to obtain those values automatically ?

Model size speed - stored in cache - stored on HDDs Cache size

Model Cache usage estimation Monitoring i/o operation on every host! Knowledge of NIC speed and HDDs speed Best way : Get this information directly form Disk Array

Model Multiple users – average bandwidth usage Divide bandwidth equally to all hosts On host, divide bandwidth equally to all users If host/user in not using its all bandwidth divide it to others Use this value as „Max Disk Array Speed” Do the same for „Max HDD's speed” value

Model Multiple users – estimating future r/w speed Don't use current speed! Use weighted mean for statistical r/w data. History time speed time s0s1s2s3 s0s1s2s3

Estimation quality tests

Estimation quality tests for multiple users

Average absolute performance estimation error: 34,8 MB/s Maximal absolute performance estimation error: 64,4 MB/s Average absolute performance estimation error: 8,2 % Maximal absolute performance estimation error: 13,9 % Average estimator response time (+ ICE) = ~1.2ms ICE – Internet Communication Engine form ZeroC

Summary and Future Works Collecting more data directly form Disk Array administration / diagnostic tools Analyzing more data – searching for patterns in Disk Array usage.

Thank You