Storage
Storage is more than choosing a capacity. I start with how the application needs to access the data, then narrow the platform and configuration options from there.
What I work on
I work with storage and solution teams to turn those platform capabilities into validated private-cloud configurations. That includes deciding which options customers should see, how they are sized, and how the storage choice fits with compute, networking and lifecycle changes.
Three ways an application reaches its data
Block
A server sees it much like a disk. It is commonly used by databases, transactional applications and virtual machines.
File
Multiple systems access shared files and folders. It fits applications that expect normal file-system behavior.
Object
Applications access the data through an API rather than a normal file system. It is common for large unstructured data sets, backups, data lakes and AI pipelines.
Why this matters to my work
The access model narrows the platform choices before capacity even enters the discussion.
Two HPE platforms, as examples
HPE Alletra Storage MP B10000
The block side of the portfolio, with file access as well. It suits workloads that expect enterprise block storage: databases, transactional systems and virtualized environments.
What I work onWhich validated configurations we offer on it, how they are sized, and how that choice fits the wider private cloud design.
HPE Alletra Storage MP X10000
Built for large unstructured data, with object and file access. That makes it relevant to data lakes, AI pipelines, backup and anything where the data itself is large and spread out.
What I work onMaking sure an AI or data platform requirement lands on a validated sizing point instead of becoming a bespoke design each time.
Choose an access model to see where it lands
The two platforms answer different questions. B10000 is for applications that expect block access; X10000 is for large unstructured data. Reading them as two sizes of one product is where storage designs go wrong.
Technical detail
Block has the most protocol variety, because these environments come with long histories and existing networks. Fibre Channel brings its own fabric; Ethernet-based transports reuse the network that is already there. That choice is often as organizational as it is technical.
Object is the model most often assumed rather than confirmed. An application written for a normal file system usually cannot be pointed at an object API without an access layer or a change to the application.
kushagragoyal.com/stack/storage