Around the Storage Block
cancel
Showing results for 
Search instead for 
Did you mean: 

Understanding the business value of HP 3PAR StoreServ

StorageExperts

 

Architecture matters. Each and every single vendor out there cannot wait to tell you why their architecture is “best.” When asked why a given technology is better than a competitor's, you will inevitably get an answer that extolls the virtues of that architecture while hearing terms like "greater resiliency, lower latency, and flexibly scalability." This tends to satisfy the question and there's either a unanimous agreement, a silent disagreement, or an open dialogue where technical opinions are exchanged. Thus begins the long process where one side does their due diligence on product investigation and the other side tries to answer every technical objection possible.

And while all this is going on, one point often gets overlooked: Why does all this matter?  Sure we’re great at using buzz words like “6-nines, 100% online, seamless, etc.” but does the point ever get addressed?

Think about this: How often do we think problems are solved through performance alone? How often is a slick management interface given priority over other more important features? How often are tradeoffs made to accommodate a “must-have” feature that may or may not actually be important? Why does this happen? It happens because vendors are great at shifting the focus to what’s important to them.

This happens way too often. Every vendor out there is very good—make that extremely good—at explaining why the one or two things they do very well should be the most important. But do they actually matter and if so, why?

Certain features are pretty easy to explain: 

  • Performance matters because the faster your solution = the faster things get done = more time to get more things done = more $$. 
  • Availability matters because downtime = nothing gets done = no $$.

If that were the only criteria then the decision-making process would be very simple: Take the fastest possible system with the highest availability rating and your problem is solved!  Right?  Wrong!

It’s always about more than just performance and resiliency—even when you don’t think it is. And that’s because something else matters! But the real question is what matters and why? 

Built for IO matters

Some vendors highlight what they do well while avoiding what they don’t by masking it with a “built for flash” slogan. HP shows you why 3PAR is the only architecture that’s built for IO up front with a design that parallelizes IO processing:

Image1.jpg 

And while those same vendors are touting “industry-standard components” or “industry-standard architectures,” 3PAR touts the engineered design that addresses IO up front. Leveraging a custom-built ASIC that works in parallel with the Intel processors, we’re able to address IO and handle integrated feature/functionality that no other architecture can touch.

Why is that important?

A shared resource model is always in a state of “constant competition.” Workload fluctuation to heavy writes and the CPU’s shift to compensate leaves fewer resources for other valuable operations. Drive failures impact those same resources. Need to replicate your data to satisfy your DR requirements? Get ready lose even more valuable processing resources. 

Shared resource models tend to upgrade more frequently, can be bound by specific types of workload, and address less of your overall workload than one 3PAR system can The HP 3PAR parallel processing model of CPU working in coordination with ASIC handles mixed workload environments with meticulous detail that no other system can match. Processing metadata and data simultaneously through CPU and ASIC while coalescing small IO and handling it as one large IO gives the HP 3PAR architecture an unmatched IO processing advantage.

The end result? 

Only 3PAR is built to address your workload and that’s irrespective of what your workload is! That’s what being built for “mixed workloads” actually means! It means you can fill your system up to the brim without fear of overrunning the performance or capacity. It also means that MORE IO is addressed because the system is designed to handle the full exhaustion of its makeup without compromise.

The system scales to a robust 460TB RAW of physical addressable space, making upgrades fewer and further between while consolidating an overwhelming amount more legacy capacity into a smaller footprint.That is how we’re able to provide a much lower dollar per GB while allowing more consumption of RAW physical capacity to data than any other vendor.

Data protection matters

Some vendors claim they don’t use RAID but then you find out they do a 23+2 stripe across a single shelf. Other vendors acknowledge they use RAID 6 but don’t give you the details of how it’s calculated or how it’s used so you would have a reasonable idea of how much RAW space you can consume but it’s still within a single shelf.

Why is that important?

Losing access to multiple drives in an enclosure causes data loss to all data in the array for systems that use this model. XtremIO can’t even sustain 2 drives failing in the same enclosure today and they’re using their own modified version of RAID 6.

Image2.jpg 

The end result?

Only 3PAR natively stripes data evenly across enclosures to deliver a shelf-level resiliency that is unmatched. We agree that enclosures themselves are not likely to fail but that’s not what this is about. Enclosure-level striping provides protection against multiple drives within the same enclosure failing—something only 3PAR can deliver!

The value is the only guaranteed 6-nines system that maintains a level of resiliency no other system can match. Staying up longer means more IO is processed uninterrupted. It’s an elimination of unnecessary fire drills or impactful outages that wreak havoc on business bottom lines. (Read more about the HP 6-Nines Guarantee Program.)

Enterprise today matters

Today, all-flash are expecting the market place to make too many tradeoffs. Shortcomings in feature-set, resiliency, and even performance are expected to be overlooked simply because they are “built for flash.”

Why is that important?

Making those concessions keeps older legacy platforms on the floor much longer than they’re needed.  The cost to power and cool those systems can be exorbitantly high especially when it’s coupled with high maintenance fees. The other side of that coin is the amount of appliances required to deliver a solution that should be included in the Flash array but is not because either the Shared Resource Model doesn’t have enough horsepower to provide those features or the product’s code simply isn’t mature enough to include them.

The end result?

3PAR StoreServ Storage is the only Tier-1 flash array with a full set of enterprise features on par with or greater than anything provided from legacy arrays today.  Replication. Application. Integration. Deduplication. Snapshots. Federation. All are fully functional and mature in the 3PAR 7450 All-Flash Array—with no compromises. 

Consider the results that a company like Sony achieved when it realized why “Enterprise Today” and “Built for IO” mattered.

By Jorge.jpgJ Maestre     @StorageMuscle    The World’s Strongest Technologist

HP Enterprise Solutions & Architectures Team

 

0 Kudos
About the Author

StorageExperts

Our team of Hewlett Packard Enterprise storage experts helps you to dive deep into relevant infrastructure topics.

Comments

Great points Jorge. The media is secondary. What is important is the architecture and how it optimizes the use of the selected media. Some of the "built for flash" architectures out there are actually  "built to rely on flash". That isn't what the HP 3PAR architecture is all about. As Jorge mentions 3PAR is an IO processing engine and much more. 3PAR has always placed and emphasis on its controller design and specifically the ASIC/CPU combination. Not only does it provide great mixed workload performance, it has the headroom to be very feature rich.