2. Tecnología Storage: situación actual 2010 Unstructured data growth unabated Data growth outpacing IT staffing IT is squeezed… Support ongoing operations Enable new growth Cost containment Total Enterprise Disk Storage Systems Shipped (EB) IDC: Worldwide Network Controller and Block-Level Storage Virtualization 2008–2012 Forecast: A Key Component in Building the Virtual Datacenter, April 2008
14. File Virtualization es: Estrategia de Storage Windows UNIX / LINUX Applications Active Inactive File Virtualization Data inspection Automated Policy Storage Decision NAS Vendor A Vendor B FC Tape Backup Clasificar datos basados en valor de negocio File Virtualization nos da: flexibilidad Mover datos sin riesgo de falla o “downtime” Políticas Inteligentes para automatizar movimiento de datos Mover datos al más apropiado equipo de storage Hacer backups SOLO en donde y CUANDO se requiera SATA
15. Debe soportar Infraestructura heterogenea Multi-vendor, multi-platform, multi-protocol NO debe limitar performance o crecimiento Más rápido que agregar unidades ó recursos físicos Debe garantizar integridad y acceso a los datos No debe ser un punto de falla Debe soportar aplicación de políticas en tiempo real Inteligencia “on-line” Elementos clave en File Virtualization Heterogeneo En Tiempo Real Intelligent File Virtualization Escalabley con Performance Integridad y Acceso
23. File Virtualization es: ahorro en Backup Move unmodified files to Tier 2 Reduces active dataset to be scanned / backed up Dramatic decrease in backup time Savings in backup media costs Reduced Tier 1 backup data set Client Tier 1 / Primary Same weekly fulls, but only 10-30% of the data F5 ARX Tier 2 / Secondary Backup Savings Example Before: 25TB x 5 weekly full copies = 125TB After: 5TB x 5 weekly full copies = 25TB Savings = (125TB – 25TB) x $0.75/GB (tape) Backup Savings: $75,000 No weekly fulls, only monthly +
24. File Virtualization: ahorros adicionales Increased aggregate utilization Power savings F5 ARX Utilization Savings Example Before: 41.7TB (25TB, 60% utilization) After: 31.3TB (25TB, 80% utilization) tiered Total Savings (tiering + utilization) = $340,000 Before Virtualization Power Savings Example Before: 41.7TB, 15k RPM SAS drives After: 31.3TB, 7200 RPM SATA drives Savings: $21,000 After Virtualization
25. File Virtualizacion es: ahorros acumulados Hard Cost Savings: Tiering + Utilization $340,000 Backup saving $75,000 Power saving $21,000 Total$436,000 Softer Cost Savings: Data mobility Lower backup/restore times Reduced backup failures Simplified administration Eliminate vendor lock-in
26. Caso Estudio:Fabricante de Tecnología A leader in designing and manufacturing technology products “ARX allowed us to perform a multi-terabyte data migration, online, without disrupting our users’ access to data, in a fraction of the time and cost to do it manually.” Director, Information Systems
27. Caso de Estudio:Institucion Financiera “F5 ARX was the only solution that offered a holistic approach to this problem. Some products can help with one piece, but only F5 ARX manages the movement of data in a way that allowed us to address the entire problem.” Managing Director
28. Caso Estudio:Fabricante de Autos “F5 ARX is a must have for IT, especially under current economic situations. It significantly simplifies unstructured data management tasks… as an IT professional for 12 years, the file virtualization technology from F5 ARX totally gets my attention” Lead Network Analyst
29. Caso Estudio:Firma Tecnología Médica FAS 3000 series FAS 6000 series R200 “With F5 ARX, I just set a policy to move the data and walk away. Users don’t know the data is moving, there is no downtime, and I don’t have to work nights or weekends to get it done.” Senior Principle IT Technologist
While stock indexes (and many IT budgets) are in decline, the amount of data is not. The fact that we’re in a recession does not mean that people can stop storing data2008 was the first year in which unstructured, file-based data drove a majority of new storage capacity in all organizations' data centers according to IDC, and they say that even with the current economic difficulties the growth in file-based information will accelerate in the coming years. By 2012, over 75% of new storage capacity shipped will be dedicated to the storage, organization, and protection of files.Over the next 3 years IT staffing will grow 1.1x while amount of information grow 4.5xResult is IT is squeezed, having to support on-going operations while enabling new growth and cost-containment initiatives What does this mean? Again, help customer see from first conversation how we can help them with this predicament
Managing storage growth top pain point by TIP. Expanding storage capacity top storage priority (IDC / Computerworld storage survey, July 2008). Opportunity in UK Mercer group. Running out of space. Do the entire sales pitch and prospect will say looks interesting but no use / no budget. Well, are you buying new capacity? Yes, sure…Cutting costs. CIO – 40% cutting and 59% freezing headcount; TIP 25% of F1000 cut by as much as 25%.Reducing recovery / restore times #3 storage priority (IDC / Computerworld storage survey, July 2008). Big problem – as amount of data grows, longer and longer to back it up (e.g. home). Only have non-business hours to do it. If something goes wrong don’t have time to start again.Mentioned up front tiering – think of migration as subset. Still very common.Vendor lock in. E.g. Allstate (insurance) – how can we free them from NTAP? Do you feel vendor taking advantage of you when it comes to price (hardware or software licenses), or maintenance on that old equipment.
What are the fundamental problems with how storage is managed today?[CLICK] First – users and applications are tightly coupled with their storage resources. Clients access file data through various static mappings or network mount points. The problem is that if you have to change things, or move data – and organizations have to move data all the time for a variety of reasons, from full-scale data migration or consolidation to storage re-provisioning – you break those static mappings.[CLICK] The second – they’re complex. Many organizations have had to manage their data growth by constantly adding more capacity, or “throwing disk at the problem.” What happens is that you end up with a sprawl of storage devices and platforms – and not always from the same vendor – that you now have to manage. A normal enterprise will have a mix of clients (Windows, Unix, Linux, etc.), protocols (CIFS, NFS, both, or multi-protocol), and storage (NetApp, EMC, Windows, etc.) and these systems don’t always integrate perfectly together. [CLICK] Inflexibility and complexity leads to inefficiency. Because it’s so difficult to move data across different devices, platforms, or vendors, traditional environments tend to be inefficient. It’s very difficult to share capacity between different resources, provision on demand, or move data around to balance utilization of. [CLICK] As a result of all this, traditional storage environments tend to be overly expensive – not only to build but also to manage. Complexity means higher OPEX costs because your staff has to manage too many, and too many different, systems, platforms, and mount points. Inflexibility means that moving data incurs IT overhead as you have to reconfigure clients with every migration, as well as lost productivity for your end users. And inefficiency means you’re buying more storage than you really need, because you can’t shift capacity to where you need it most.
Lower CAPEX – use cheaper storage (tiering) and use less of it (utilization). Also less backup media.Lower OPEX – more efficient power consumption (tiering) esp. to dedup, eliminate high support costs to maintain end-of-life gear, automate manual tasks so staff can focus on other issuesOther benefits:Data mobilityNo downtime, no impact to appsReduced migration timesFree up IT resourcesGreater flexibility to integrate new storage technologies Added bonus Greater leverage when negotiating with vendorsOther softer benefits:Transparent cross-vendor capacity expansionTransparent cross-vendor migrationTrue vendor choiceReduction in backup/restore timesReduced backup failures
Well, one customer described it to me like this.He shows this picture and he says this was my storage environment before tiering (ILM)…It was a mess. Stuff all over the place and I couldn’t find anything.
This is my environment after tiering (ILM). Everything where it should be.Making more efficient use of space.I thought it was a good analogy, but I thought we needed to add one more step.
Now I want to switch gears here and talk about another application of data mobility…Something we have found consistently in working with hundreds of different customers over several years is that very few have a good handle on their unstructured data. In the file world it is difficult for admins to get a holistic picture things like type of data they have, how old it is, how frequently it’s being used, by whom… Results in treating all data same. Put it on same storage devices. Back it up in same way. Reality is not all data is created equalNot all data has same value to the business (e.g. mp3, jpg)Some data hasn’t been touched for years yet it still sits on fastest servers and gets backed up every week.And some data might require better performance than our generic vanilla storage tier can provide F5 has software app called Data Manager which gives you this visibility (Diagnosis)ARX can do something about it. (The cure)
How?Simple example. Consider just creating 2 tiers. (Can be more).Set policies on ARX to tell it where to place data.Data important to businessData active.Requires performance. Usually end up with a lot, lot more stuff on tier 2Interestingly, today all tier 1.One customer described it to me like this… (file storage like my garage. Before F5 lights were off. No idea what was in there. DM turned lights on and realized what a mess I had. ARX like home makeover crew – after installed garage ad for container store / IKEA)
We’ve taken a look at how this works, now let’s drill into some the benefits in a bit more detail.As we move data off of tier 1 primary storage, that expensive disk capacity is freed up.The majority of data is moved to less expensive tier 2. No administrator intervention is required throughout this process, and users and applications are not affected in any way.Added benefit that Tier 1 is less heavily loaded.What sort of savings are we talking here? Well, they can be sizeable. If we took a 25TB example – 25TB of tier 1 disk capacity – and are able to move 80%, or 20TB, of that data to a more cost effective tier 2, the cost savings amount to around $180,000. But that’s only a part of the story.
By moving 70-90% of the data to tier 2 we also dramatically reduce the amount of data that needs to be backed up regularly. If the data has not changed in 6 months, you don’t need to keep backing it up every week. Files on Tier 2 have already been backed up on Tier 1, so they can be backed up less frequently on Tier 2. Tiering ensures we’re only backing up the data which is changing, and backing up inactive data far less frequentlyE.g. a typical tier 1 backup schedule might be nightly incremental backups and weekly full backups, with additional monthly / quarterly / or yearly full backups. If we have 25TB on Tier 1 and retain 5 weekly full copies of the data, that requires 125TB of tape media just for the weekly full backups. Now, if we move data older than 3 months to tier 2 (say 80% of the data), and we continue to backup Tier 1 on a weekly basis and Tier 2 is only backed up on a monthly basis, tier 1 is only 5 TB and those same weekly full backups only take 25TB. That’s a savings of 100TB of tape media. The fact that we have decreased the backup data set by 70-90% also means that the time taken to complete a full backup drops significantly (will also decrease scan times on tier 1). Back to our earlier example, a full backup of tier 1 would be 5TB rather than 25TB – should be around 1/5th the time. (Further reduction in time since the backup agent need only scan Tier 1).
Data growth rates ranging from 60 to 80% a yearThe IT organization knew the backup environment had reached its limits. Monthly backups had grown to almost 100 terabytes (TB), even though, on average, only around 15% of the data was actually currently relevant. In addition, many terabytes of e-mail archive (.PST file) data was backed up every day – much of it unchanging.Digging deeper into the problem, the IT organization discovered that home directories were consuming almost 40TB of tier 1 (NAS) and growing rapidly. A significant amount of the capacity in the home directories was devoted to .PST files, and just 5% of all the files were less than 90 days old.“With F5 Acopia we set policies that reduce our active data set to files less than 90 days old. Just pulling out the old .PST files cut our backup environment from 400TB to 4TB. The differential is huge.”Wanted to move away from EMC, so brought in NetApp to move this less active data (80%) of the data to it. Cap EMC investment and phase out as depreciated.
50% p.a. growthThey were actually on a Windows Storage Server platform.We have 12 hour backup windows, but backups were taking 18 hours to complete.“We know we cannot stop data growth, so we had to come up with a plan to dealwith it,” Chen says. “We decide to move to NetApp NAS and incorporate ILMtechnology so we could separate the data that we really cared about from the datathat was less critical to us.”Actually have 3 tiers; data <3 months on tier 1 (FC), 3-18 months tier 2 (SATA), >18 months on tier 3 (SATA). Backup tier 1 with daily incrementals, weekly fulls, tier 2 weekly fulls and tier 3 monthly fulls. They also use ASIS to achieve further efficiencies on their storage, but can use ARX to determine what data they want deduped.(Also mp3 policy – push to share with no space).Backup windows from 18 hours to 1.5 hours.
Huge amounts of storage. MRI, telemetry, finite analysis, media, etc.Massive growth. 0.5TB in 2001 to >100TB today; >100 million files.Managing a complex NAS environment with >60% growth (in addition to NetApp there were Windows Servers and Unix/Linux Servers)Wanted File virtualization to simplify complexity, minimize disruption and optimize backups, but had to work in multiprotocol environment (unique to ARX)Benefits:Improved utilization ~90%Reduced backup times . Smaller physical volumes with multiple backup streams in parallel. ~20x improvementEasier administration. Set migration policies and walk away; no more nights / weekends . Tasks that took 4-6 hours now take 30 - 45 minutes.Also doing tiering and running ASIS on tier 2 volumes.