Delphix Platform Overview


Published on

Cos'è Delphix....?
Quali vantaggi porta un'architettura basata su un Database Agile Management...?
Alcune informazioni utili............

Published in: Technology
  • Be the first to comment

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide
  • Standard APIs - Non-disruptiveDB aware compression and filtering – more efficientAny storage – avoid lock-in
  • Database upgrades can be very taxing both from an operations and systems perspectiveUpgrades often require repeat access to database copies from production, impacting production DBAs and app teamsMost organizations will run through a full cycle of development and quality assurance before rolling out upgrades to avoid production downtimeThey will need to do this for each application for a given database, because applications behave differently with database changesOrganizations often will have rehearsal environments for testing an upgrade process before a production rolloutThey may also need copies for training environments to train users or administrators on new features or functionsDuring the course of an upgrade, organizations should get repeated refreshes or recent copies of production databases to ensure no changes have been made in production that will break during an upgrade processDelphix can dramatically simplify all phases of an upgrade programVDBs (and sources) can be separately upgradedDelphix dramatically reduces the storage required for the upgrade processBy shortening the cycles for an upgrade, Delphix also improves the use of database serversDelphix can also improve the quality of an upgradeFor both applications and databases, there are a large number of potential configuration changes that can be tested during an upgrade to maximize a successful outcomeDelphix can quickly and easily create a number of A/B virtual sandboxes to test configuration A versus configuration B versus configuration C….which can result in a more successful outcome.
  • Performance of virtual copies relative to physical copies can vary by workload. Several architectural elements mitigate physical vs. virtual performance. These include:Read and write cachesAbility to augment cache benefit with SSD on Delphix ServerOn-the fly compression, de-compression
  • Source host has an HTTP connection (ports 21000-22000) from each source host to theDelphix ServerTarget host has an NFS connection (ports 22000-23000) to Delphix Server – used when provisioning VDBsDelphix connection to source and target host is over SSH (port 22)Delphix connection to Oracle on the source and target is over a JDBC connection using Oracle Listener ports (for initial discovery)Delphix Server has DNS (port 53) configured and NTP (port 123) configuredDelphix Server has LDAP (port 636) configured and SMTP (port 25) configuredDelphix Server has NFS (port 111, 1110. 2049, 4095) configuredDelphix Server has HTTP (port 80) and HTTPS (port 443) configuredIf replicating or backing up the Delphix server, configure ports 21000–24000 as part of the profile   
  • Delphix Platform Overview

    1. 1. AGILE DATA MANAGEMENT Simplify 10x, Accelerate 5x with Database Virtualization
    2. 2. Delphix Agile Data Platform Delphix transforms the economics of database and application management through: – Flexibility • Any storage, deployment model, (database) – Elasticity • Dynamically address app dev demands – Self service • De-couple app dev dependencies on IT – Cloud readiness • Unblock data tier anchored to legacy models – Agility • Drive app dev project output gains
    3. 3. Initial Sync With Production (or Standby) Production Oracle 9i, 10g & 11g on Linux and Unix • One time full copy using standard APIs • Intelligent compression (75% reduction) A D G J B E H K C F I L • Supports any storage Full Delphix Virtual Appliance A B C D E F G H I J K L
    4. 4. Ongoing Sync - TimeFlow Production • Combines continuous log collection • With incremental backups • To build a continuous window or “TimeFlow” of changes TimeFlow Full Incr A B C D E F G H I J K L E Blocks A D G J B E H K C F I L H /Logs D A E F J Logs Incr Incr A E E H A H H A EE E EE E EE E
    5. 5. Virtual Copy Provisioning (or Refresh / Rollback) • Select any point in TimeFlow Dev QA • Appropriate blocks are shared (not copied!) • Copies of TB sized DBs consume a few GBs D G J E H K A D G B E H F C A B C D E F G H I J K L L Incr A Blocks Full I Logs Incr F I Incr A D E D D F H J EE E EE E EE E …..
    6. 6. Virtual Copy Provisioning (or Refresh / Rollback) • Re-parameterizes, starts, mounts copies Dev QA • Executes recovery and rolls forward with logs • Copies of TB sized DBs fully ready in minutes D G J E H K A D G B E H F C A B C D E F G H I J K L L Incr A Blocks Full I Logs Incr F I Incr A D E D D F H J EE E EE E EE E …..
    7. 7. Upgrades 1 2 • • • 4 Prepare Oracle Home Backup or Restore (Retry) 3 Upgrade Database Validate High cost of creating copies to test compatibility with various app configurations Significant reliance on production teams impacts other projects Need for frequent refreshes and iterative testing introduces huge operational burden 1 * 2 * 3 * 4 • 2 – 4X faster upgrades: eliminate operational, hardware bottlenecks • Reduced risk: more environments, latest data, greater repeatability • Lower project costs: storage, bandwidth, labor >> Strictly Confidential
    8. 8. Agile Development Project team B Project team A Project team C • • • Days to weeks of setup, load, configuration, tear down time prevent short cycles Large number of environments for parallelization introduces a cost barrier Significant effort around integrated testing and slow recovery further limit adoption Project team A Project team B Project team C Project team D Project team E • Scale without rising costs: copies per developer or project (90% lower storage costs) • Maximize development time: setup, tear down effort cut to minutes (95% reduction) • Simplify integrated testing: create, refresh, recover copies to exact same point in time
    9. 9. Master Data Management X • • • # of synchronized copies needed Requires coordination across multiple production teams to restore from specific backups Followed by tedious and manual log application to synchronize copies Generally adds weeks of delay over and above standard refresh processes 12:21:05AM • Synchronize down to the second through automated log shipping and application • Flexibility of self service synchronization for developers by time or transaction ID in minutes • Accelerate MDM projects by up to 70% by eliminating both labor and coordination time
    10. 10. Backup and Recovery 2 to 3X storage for 1 week of retention 7X storage for 1 week of retention • • • Improving RPO, RTO requires more full backups, more frequent backups  high storage costs Conversely, optimizing for storage costs means fewer recovery points and slower recovery Recovery is further slowed by standard provisioning processes < 0.5X storage supports weeks of retention and all virtual copies • 90%+ cost reduction: incremental forever updates eliminate RPO, RTO vs. storage tradeoff • 98% faster recovery: with integrated log shipping and orchestration of DB layer tasks • V2P: (Virtual to Physical) functionality extends backup protection and savings to production
    11. 11. Data Center Migration and Consolidation App Tier Host / Server Tier DB / Storage Tier High CapEx, OpEx costs for test infrastructure Phased / long term projects High risk and variability in both cost and effort App Tier Host / Server Tier DB / Storage Tier 90% reduction in storage cost for copies 95% faster critical path tasks 30% to 70% project acceleration
    12. 12. Accelerate Migrations by 30 – 70% Delphix enables schedule, risk, and quality benefits of parallelized migration testing with lower costs Agility and cost benefits can be sustained post migration by using Delphix for app development • On-demand copies that occupy less than 10% storage of physical copies • 95%+ lower effort to provision, refresh, rollback for fluid migration rehearsals • Bandwidth efficient sync over WAN through “incremental forever” updates • V2P (Virtual to Physical) after migration validation to the target prod server • Adapters to automate platform or DB migration / upgrade processes
    13. 13. Competition Delphix Agile Data Platform Benefits Perceived Competitors By Category • Storage vendor lock-in; lower efficiency (no DB awareness) • High risk to app quality; lower efficiency (fixed cost per subset) • Storage clones • Partial data subsets Storage Savings for Backups • De-duplication • Storage vendor lock-in; lower efficiency (no DB awareness) Downtime Reduction / Forensics • • RPO, RTO vs. storage cost tradeoff • Traditional backup (flash / disk / tape) Replication for DR Provisioning and Refresh Labor Savings • Storage clones • No orchestration of DB layer tasks cross team dependencies still bottleneck app dev • VMWare Data Director • Complimentary – helps with app and DB server tier tasks (data tier is the bigger bottleneck) • VMWare Data Director • Complimentary – helps with app and DB server tier tasks (data tier is the bigger bottleneck) Storage Savings for Copies App Dev Project Output Multiplier • No protection against logical failures
    14. 14. Storage Clones ROI Source / Capability Storage Cloning Tools Delphix One hour with high reliance on manual error prone steps – full repetition even for a refresh End to end automation, provision with 3 clicks in minutes; refresh with one click or automatic based on policy Provisioning granularity No continuous sync – limited to snapshots of “Golden master” Down to the second / SCN provisioning via integrated log shipping and archiving – across multiple databases Self Service Enablement Process requires more custom scripts to provide full provisioning processes, with no simplified “refresh” 3 clicks, simple GUI designed optimized for app dev and analyst self service workflow Initial load performance Block level storage copy may be faster Supports 10GbE N Y Limited- applies to taking snapshots Y General purpose filesystem - no native compression Shared footprint for duplicates in the same filesystem Database specific filesystem, Intelligent compression, filtering, shared footprint and sync enable greater efficiency Y N NetApp lock-in Vendor agnostic Masking tool integration N Y Resource management / QOS Limited: By File system - requires additional golden copies Y - Database level granularity QOS and resource management N Y OpEx Provisioning & refresh effort Self-service capability Policy based operations CapEx Storage efficiency Unnecessary blocks copied Storage flexibility Risk Reduction RBAC & auditing per copy
    15. 15. Data Guard and Flashback Category Data Guard with Flashback Delphix Storage Up to 2.4*(Source DB) per Forensic Copy Up to 0.4*(Source DB) per Forensic Copy; 80% savings. Network All Blocks and all Redo shipped over network per Forensic Copy. Blocks and Redo shipped once over network for all Copies. Time to Restore Variable. Restore time can be very long; Redo time is bounded by flashback Interval Can require flashback, and redo application and be fully manual. Constant. Restore is almost instantaneous; Redo time is similar. Reuse Sync & Catch-up Time to Revert to Standby is a function of disconnection time. Simultaneously provision and re-provision VDBs at multiple points in time in a few clicks. Delphix continues to keep the dSource in sync with the Primary, even when multiple VDBs are provisioned and actively in use.
    16. 16. High Availability Non-Prod, Offsite Non-Prod, Onsite Database Servers 1 MP Redundant TCP Network •1 •2 MP 3 Replication 2 NDMP Backup 3 4 MP SAN SAN •4 SAN MP SAN • Network multipath Support (NICs) Zoning: replicate within, across sites Backup, archive .VMDK to 3rd parties SAN multipath support
    17. 17. Performance VDB1 VDB2 VDB3 NFS 1 VDB4 2 3 4 •1 •2 •3 •4 Global “Hot” Read Cache Global Write Log SSD Flash Cache On-the-fly Compression, De-compression
    18. 18. Connectivity Source Server Target Server HTTP SSH JDBC SSH JDBC Compressed Databases NFS
    19. 19. System Requirements Delphix Server – Virtualization Platform: VMware (ESX Server 3.x, vSphere 4.x, vSphere 5.x) – VM Guest Container Profile: Solaris 10 64-bit – Virtual CPUs: 4 vCPUs recommended, (2 vCPUs minimum) – Memory: 16 GB (minimum), add more if indicated Source and Target Servers – Oracle, Oracle 10g R2, Oracle 11g R1, or Oracle 11g R2 – Solaris, RedHat, OEL, SUSE, AIX, HPUX
    20. 20. Thank You