LocalData

I think it will become table stakes. If a solution isn't scale out, it won't make the short list.

even if the buyers don't know exactly what scale out is, they'll know it provides certain characteristics they care about

@andriven agree, the question is will true scale out be table stakes or some modicum of it

Scale-out will be a fundamental requirement for secondary storage in 5 years (Backup, Test/Dev, Analytics)

At the edge, everything is converging as endpoints get more powerful (look at the power of an iPhone). In the datacenter, everything has to scale out.

Most enterprise data will be stored, protected and accessed via scale-out systems in 5 years (trend already underway)

will be required for anything of significant size, have to have the ability to lose a node or nodes without impact

Scale out in the form of hyperconverged infrastructure is already upending the industry. Classic large storage margins and long services engagements are a huge opportunity for new companies to productize the traditional complexity + long standups

scale out is going to turn the traditional backup industry on its head. That will be the easy entry into the enterprise.

I think that scale out will be the driver for "multi-cloud" where the world will spin back to the datacenter in a new form factor cloud ops will be the new IT operating model

Software apps will leverage a horizontal stack (a new capability) that will automate lots of non-differentiated stuff and move the value activities up the stack

@johna_white Couldn't agree more (obviously). In the "solution searching for a problem", scale out can impact faster in areas where customers don't like their current products. Backup is classically one of those areas - no one likes what they're using.

If you think about scale out on the macro level in sites instead of hardware objects we will see a new demand on the network that will require further innovation past just 5G.

I think scale out has confused the industry. if you can scale out backup infinitely is it backup or is it archive

@johna_white Indeed, IoT will drive "scale out like" solutions, though I'm sure we'll know them by another name.

@csharney I'd posit that the difference between backup and archive is RTO. Some day if RTO is free there's no diiference between backup and archive. Not the case today though...

@csharney Backup vs. archive is a conversation that needs well defined terms in order to be useful.

@csharney Difference between backup and archive is merely the retention policy and whether you choose to delete the data off primary storage. 😀

@fdmts In some ways, I think of IoT just as "the source of the growing data tidal wave". I almost don't care what the data is...there will just be so much of it that we can't cope without some major infrastructure paradigm shifts (bingo!).

@andriven I think of archive as being associated with immutable, primary data and backup being a versioned thing.

@csharney Scaling out secondary storage has been the response to growing silos of primary storage that have often been constrained for performance, management, or technology lifecycle reasons.

@dinisco I'm truly unsure if "scale out" will become enough of a brand name for most customers to look for it. The capabilities it creates definitely will be required + table stakes though if only because they align with all the public cloud promises + hyp

Compute, storage, networking...all-scaling out. It's the applications man!

Heavily adopted in HPC, partly adopted in primary storage. Hyperconverged is an example of scaleout eating primary storage.

Compute farms / HPC already strive for scale-out.

HPC, Networking

@steve_pao agree for strive to be scale out. Not all do

compute, storage, backup, and HCI

but from a mindshare perspective, I'd argue people ask about/see the need for scaleout already in Networking, Primary Storage, and Compute Farms/HPC. It's on the adoption/mindshare curve for Backup.

Diaster Recovery is interesting as it's a combo of backup/storage/networking/etc.

I disagree about HPC, unless we're talking about Spark and similar architectures. Batch HPC is anything but scale out when we include network and storage.

block storage hasn't really happened save a few exceptions

In #publiccloud you presume that scale out is happening. I don't think any public cloud provider isn't using scale out tech under the covers in order to provide elasticity.

administrators don't have to interact with scale out components, just need to understand there are a "fleet" of resources that can scale out as needed

Not totally clear on the question. Do you mean full-on strawman legacy vs. "modern?"

Depends on from whose perspective.

In the #datacenter you still often have to consciously choose between scale out and non-scale out architectures weighing their pros and cons.

Show me a public cloud that is "scale-out" in terms of logical namespace. Cloud is bag of bolts that still need software to interconnect it.

this is the argument for leveraging native cloud services vs. traditional

I don't know that scale out is any different in the private datacenter than the public cloud, other than some one else is managing the infrastructure

In the data center we use it run our business and in public cloud it is their business. Completely different focus on the word "scale"

In the spirit of people, process, technology, the technology has to support different people & process.

#datacenter systems should be designed with public cloud principles in mind. Otherwise they're boat-anchors.

where often non-scale out products can be cheaper on a short term basis as their design cost has long been amortized. Think Networker sold for 99% off at times.

In the #datacenter, it's still your job to scale it. The #publiccloud does it as a business

In the @IgneousIO view, we evangelize the need to deliver on-premises infrastructure as-a-Service and have strictly separated our data plane from our management plane.

@fdmts 95% of apps running in data centers today don't require infinite scale. Why spend the money to shoot for an impossible dream?

achieving the scale and elasticity offered by most public cloud providers in the DC often requires underlying scale-out infrastructure, issue is having hw resources at the ready without over spending

As a developer today, you can make assumptions that the #publiccloud you are using can scale as you need it. You generally can't make that assumption in the #datacenter.

@johna_white Infinite scale isn't necessarily required, but the scaling increments may be very different - particularly as you deal with fault domains.

would also say that scaleout in #publiccloud is often designed/tested to scale far further with more automation than in the #datacenter . Just a matter of economics...

@johna_white Infinite scalability isn't necessarily the goal of cloud adoption. For me, it's agility and the potential to move / re-use systems.

Most people don't need the ability to massively scale but ask a a developer if he prefers always having resources available when they need or for as long or short as they need it or if they prefer requesting resources from IT and waiting...