NetAppChat

A6: IOPS

The race to 0us is on. It's all about that latency. Period.

A6: @datacenterdude agreed. But my question is how many people and apps really need near-zero latency?

@dancbarber Is flash fundamentally changing how apps are designed?

A6 : Flash is being leveraged for high-performance applications that speed response time. This is a great example of a practical application of flash. #boom

Storage bottlenecks have a high cost associated with them. Eliminating the bottle neck means more efficient servers, which equates to less of them. More importantly less software licenses, this is where the real saving take place.

@ajbergh definitely.

You wouldn't be surprised. Some of the biggest names that host data for the consumer digital products you use every day run on EF550/560, and that was a core requirement of all of them.

@ajbergh Flash is showing app developers the potential they have. Case in point: what if a major retailer could get realtime info on every purchase a customer made right at the Point-of-Sale device?

A6: As this customer says: flash is helping his team consider workloads not possible before. http://www.netapp.co...

@ajbergh Now me, the retailer, can offer relevant services instead of "would you like to sign up for a rewards card? Oh wait, you already have one :)

A6: Server-side flash avoids bottlenecks for deduplication.

A6 : Gee, I know of a SAN that backs up enterprise applications in seconds and restore them in minutes... #NTAP

@DataDefiner Flash, especially the way NetApp has designed their FlashRay, uses global inline deduplication, byte-granular inline compression, variable length block layout and adaptable performance.

@DataDefiner do you see a multi-level caching strategy? VM-HBA-Host-Switch-Controller-Disk?

@GildaFoss A6: Hey @GildaFoss That sounds like a data platform I know, too! #CV

Mark nailed it. The biggest cost-savings is in less hw required, which equates to less licenses and support contracts needed. Doing more with less always saves. Been NetApp's M.O. for a long time. "Go Further, Faster"

@dancbarber A6: More than just IOPS!

@dancbarber Absolutely! All the time with CommVault.

@DataDefiner Love me some @CommVault!

My favorite is hearing about customers who go from overnight queries to over-lunch queries with #netappflash

@GildaFoss Flash also works great with source-side dedupe, inline compression, etc. I recommend server-side flash and/or AFAs all the time!

Anyone that thinks the whole datacenter will be all-flash all-the-time for all-the-things is silly in the head. There will always be a need for "cheap & deep" for WORM, archival, cold storage, and off-site long-term retention.

(cont) and that will still constitute a MAJORITY of total data in any enterprise.

A8 : The most successful enterprise storage strategy will continue to be one that uses both flash & hard disk. Over the next 5 yrs, hard disks won't begin to approach flash perf, & flash won't begin to approach the capacity needed. #justfacts

@datacenterdude agreed. there will always bee the cheap & deep

Drop an All Flash FAS in a cDOT cluster...

@GildaFoss well said.

A8: One word: Data mgmt. Archiving. Not everything needs to be on tier-0/1!

A8 : Bottom line is that the all-flash array that have no integration with existing hybrid flash/disk arrays, will be the hot rod in the garage of IT. Fun to tinker with, but not the reliable storage workhorse IT needs. #period

Flash as a caching tier with low cost SATA, near-line SAS for capacity, all flash array on pre with capacity in the cloud

A8: And don't forget @AWS and @Azure -- data must be searchable, recoverable, etc.

@DataDefiner and ownership-able

A8: Performance disk is in trouble, but capacity disk will have a big $/GB advantage over even the cheapest for of flash for a while. Hybrid architectures (flash+disk), whether on-prem or leveraging the cloud, are the new normal.

A8: Hybrid approaches - Flash as a tier or a cache to traditional media will be the new "normal" in the near term

A8 : Enterprise storage continues to move rapidly to a hybrid model, where models are applied both within & beyond the enterprise's traditional boundary between hardware it owns & manages.

@DataDefiner Good point, Brian. I still hold firm that DRaaS will become one of the first major uses of public cloud from the enterprise. From a simple capacity play. Hell, I do it at home!

@GildaFoss Everything is moving toward hybrid. There isn't a one-size-fits-all SAN, nor is there cloud, compute, network, application.

Driving the point home : The Future of All-Flash Arrays is Not All Flash.

@GildaFoss or at least not everything on an AFA

Maybe as @mo6020 said an #AFF within a larger #cDOT cluster

@datacenterdude Yep, and long-term cold storage.

All flash for storage simply doesn't scale... data has tiers and so will storage. With Software Defined Storage #SDS, customers can look creating storage management policies and automating data movement in line with "data tiers"...

A4: Faster data access is key; #flash is certainly helping that.

The top three things I hear: a) Ease of setup/maintenance b) connecting to host environments (i.e. proper zoning/config) and c) Staff doesn't scale as well.

@datacenterdude Agree on the staff part. When personnel is one of the biggests costs, a scalable SAN is very important to business effectiveness.

.@ripw5280 I'm not SAN expert but I assume that it is important to pay attention to the fundamentals of SQL Server

(cont) These are not new problems. As much as we want to believe Flash will be the end-all, be-all for SAN, the reality is many of the core problems still exist. SAN is still SAN.

Complexity of data migrations

@arianagradow it is certainly fundamental to understand your workloads. Too many people just say "put it on flash and don't worry about it."

@datacenterdude I agree. Flash deals with mechanical latency and it can help with copy creep but you still need a management stack

meeting backup windows and backup for remote sites

A4: Challenges are lack of granular data recovery; no easy way to manage compliance; data migration too costly.

A4: 1. Ease of use. 2. Updated infrastructure to use advanced features. 3. Training people to use and understand the new technologies.

A4 : Top 3? Biz-critical DB’s & apps running by virtually eliminating storage downtime, improving ops perf (make better-informed decisions), increase the efficiency of DB’s+ apps & reducing IT mgmt workloads, & high perf with high productivity

R Cox, how much does bandwidth play into this?

@dvellante @datacenterdude Not mention personnel and operational latency—I think we could solve a lot of problems that way!

@dvellante Could you please expand on data migrations? Have you looked as thin provisioning and reclamation? #NetAppChat

bandwidth and dedup are important