What you’ll learn in this tip: In this cloud storage review find out how industry analyst Marc Staimer defines
cloud storage and read his list of cloud storage requirements. Staimer also explores the cloud washing phenomenon created by marketers and vendors.
All the hype and noise surrounding clouds and cloud storage is causing quite a bit of confusion. Some vendors like it that way because it allows them to obfuscate cloud storage definitions, cloud storage capabilities and cloud storage value propositions. This sort of marketing spin is called cloud washing, and it slows the adoption of true cloud storage offerings.
Cloud storage requirements
True cloud storage is storage that’s accessed directly across the Internet via Web services APIs. These APIs utilize variations of the REST or SOAP interface that will eventually be standardized and known as the Cloud Data Management Interface (CDMI). In my opinion, users can define cloud storage simply by knowing specific requirements. What follows is a list of technical requirements that a cloud storage offering should meet.
- Massive scalability: always increases positively as storage nodes are added
- Dozens of petabytes to exabytes, and even zettabytes
- Tens of billions to trillions of objects and/or files
- Double-digit GBps to TBps of throughput
- Not tied to geographic location (geographically aware)
- Objects/files aren't storage location dependent
- Move objects/files based on policy such as user response times
- Based on off-the-shelf components to minimize total cost of ownership (TCO)
- X86 servers
- SAS, near-line SAS (NL-SAS) or SATA hard disk drives
- Multi-level cell (MLC), enterprise MLC (eMLC) or single-level cell (SLC) solid-state drives
- Standard 1 Gb, 10 Gb and 40 Gb Ethernet; no requirement for Data Center Bridging, also known as Converged Enhanced Ethernet, which is more costly
- Secure multi-tenancy
- No unauthorized user or employee can ever read/write someone else’s data
- AES 256-bit encryption, National Institute of Standards and Technology (NIST) FIPS 140-2 certified, erasure codes, access control lists (ACLs) and even automated rotating passwords
- High degree of data self-healing
- Extremely low probabilities of data ever being lost, corrupted or silently corrupted
- Enduring data resiliency/permanence
- Accessible online for years or decades, searchable and capable of locking down data with immutability via WORM
- Has to outlast many multiple technology refresh cycles
- Allocation on-demand: policy based and user transparent
- Capacity, performance and location
- Billed or licensed per usage with true “pay-as-you-go”
- Charged on usable storage vs. traditionally raw storage
- Also known as “pay-by-the-drink”
- Application-agnostic with transparency to all applications
- Structured or unstructured
- With REST, SOAP and/or CDMI interface
- Nondisruptive tech refresh; i.e., no scheduled downtime ever
It's important to note that these aren't optional requirements for public cloud storage. Secure multi-tenancy means never having to say you’re sorry to a customer for a breach. Self-healing features ensure data is readable and searchable when required. Enduring data resilience/permanence is a necessity to have a compelling value proposition over vaulted tapes. A pay-as-you-go model allows users to pay for only what they're using, when they're using it, so they never have to pay for unused or unusable storage. Application transparency simplifies and accelerates cloud storage adoption whereas a disruptive traditional storage tech refresh is a non-starter for service businesses that can never be down.
Overview: Public cloud vs. private cloud vs. hybrid cloud
Let’s look at public vs. private vs. hybrid cloud technologies. Public cloud storage has users and applications connect to the storage across the Internet into a service provider’s data center(s). On the other hand, private cloud storage has the cloud storage residing in the user’s data center(s) or in a Infrastructure as a Service (IaaS) data center(s) that is rented/outsourced and not shared with anyone else. A hybrid cloud model leverages the best of both public and private clouds in a wide range of custom implementations. Hybrid cloud storage can be set up to put everyday stored data in a private data center, while older or archive data is automatically migrated to the public cloud. A hybrid cloud can also leverage public storage clouds for disaster recovery scenarios. The range of options is only limited by the imagination of the administrators.
Regrettably, these descriptions don’t offer many distinctions between public and private clouds, which can lead to considerable confusion. Some people will tell you that NAS, scale-out NAS, scale-out SAN and iSCSI SAN storage are types of cloud storage.
Currently, the only storage technology that currently meets cloud storage requirements is object storage. But this doesn't mean that all object storage systems, such as Open Stack, for example, meet all cloud storage requirements. Finally, not meeting the listed requirements isn’t a mark against some storage systems, it just means they aren’t a true cloud storage offering.
BIO: Marc Staimer is founder and senior analyst at Dragon Slayer Consulting in Beaverton, Ore. The consulting practice of more than 13+ years has focused on the areas of strategic planning, product development and market development. He can be reached at firstname.lastname@example.org.