Unintellectual Thoughts

Emptying the dresser drawer of my mind.

  • When will all-flash protection storage become the “hot new thing”? To deal with the increased scale of primary storage capacity and more demanding customer SLAs, the industry is moving from traditional tar/dump backups to versioned replication. Thus, protection storage needs to support higher performance data ingest and instant access recovery. It seems plausible that protection storage will follow the primary storage path: from disk-only to caching/tiering with flash to all-flash (with a cloud-tier for long-term retention).
  • When will custom hardware come back? The industry has pivoted so hard to commodity components, it feels like the pendulum has to swing back. Will hyper-converged infrastructure drive that shift? After all, where better to go custom than inside a completely integrated end-to-end environment (as with the mainframe)?
  • Are job candidates the biggest winners in Open Source? Companies continue to struggle to make money in Open Source. Whether the monetization strategy is services, consulting, management & orchestration, or upsell, it’s been a tough road for Open Source companies. On the other hand, Open Source contributions are like an artist’s portfolio for an engineer– far more useful than a resume. Even better, if you can become influential in Open Source, you can raise your profile with prospective employers.
  • When will NAS evolve (or will it)? It’s been decades since NAS first made it easy for users to consolidate their files and collaborate with their peers in the same site. Since then, the world has evolved from being site-centric (LAN) to global-centric (WAN). Despite all the attempts – Wide-Area File Services (WAFS), WAN accelerators, sync and share – files still seem constrained by the LAN. Will NAS stay grounded or expand beyond the LAN? Or will object storage will simply be the evolution for unstructured data storage and collaboration?
  • What’s the future of IT management? Analytics. We’ve spent decades building element managers, aggregated managers, reporting tools, ticketing systems, processes, and layers of support organizations to diagnose and analyze problems. As infrastructure commoditizes, we should be able to standardize telemetry. From that telemetry, we can advise customers on what to do before anything goes wrong. If companies like EMC can make technology that reliably stores ExaBytes of storage around the world, we should be able to make technology to enable customers to not have to babysit those systems.
  • Will Non-Volatile Memory be the disruption that we thought Flash would be? Flash didn’t disrupt the storage industry; it was a media shift that the major platforms/vendors have navigated. (Flash did disrupt the disk drive industry.) The non-volatile memory technologies, however, could be more disruptive. The latency is so small that the overhead of communicating with a storage array exceeds that of the media. In other words, it will take longer to talk to the storage array than it will to extract data from the media. To optimize performance, applications may learn to write to local Non-Volatile Memory, shifting storage out of the primary I/O path. Maybe that will be the disruption we’ve all being talking about?
  • What happens when storage and storage services commoditize? The general consensus is that the commoditization of IT infrastructure is well under way. Most people feel the same about storage and storage services (e.g. replication, data protection, etc.) As commoditization happens, customers will choose products based on cost of purchase and management. As an infrastructure vendor, the question will be – how do we add value? One camp believes that the value will move to management and orchestration. I’m skeptical. Commoditization will lead to storage and services being embedded (e.g. converged/hyper-converged) and implicitly managed. Thus, I think there will be two paths to success. One path involves becoming a premier converged/hyperconverged player. The second revolves around helping customers understand and manage their data – on and off-premises. This means security, forensics, compliance, and helping users find what they need when they need it. Successful vendors will either deliver the end-to-end infrastructure or insight into the data. If you do both… then you’ve really got something. You can guess where I’d like Dell EMC to go.

I also wonder about whether software engineering jobs are following the path of manufacturing jobs, whether software-defined XYZ is a bunch of hooey, the future of networking, whether any of these big-data unicorns has a shot at success, and why people are so hysterical about containers. But we’ll save those incoherent thoughts for another time.

-Stephen Manley @makitadremel

One Reply to “Unintellectual Thoughts”

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s