Why are Large Form Factor (LFF) disks still fairly prevalent?

Question

Small Form Factor (SFF) / 2.5" disks seem to have become more popular than LFF disks now due to them being preferable over LFF disks in many scenarios (lower power consumption, higher density, etc). However, LFF disks still seem to feature in major vendor's offerings (take the recently released Gen9 series of HP servers as an example).

Looking at the pricing of the disks, in most of the lower (sub-500GB) capacities, there seems to be little price difference these days. That begs the question, why are they still popular enough for vendors to feel it's worth investing in supporting them in their latest products? Is it purely because the LFF form factor disks are available into higher capacities than SFF disks, or are there other reasons why they are still popular?

Underlying this is the fact I'm trying to understand what objective justification there is for spec'ing out a modern server with LFF cages/disks over SSF. What scenario/requirements might mean LFF would be the preferred choice? Would you only really do this if you needed large, multi-terabyte disks at a sensible cost, or are there other reasons?

Solution 1

Use 2.5" disks for enterprise SAS workloads and 3.5" for bulk and high-capacity storage.

You've answered your own question. Buy the right type of server for your anticipated workload. If you need high performance drives, optimize for that. If you need a lot of storage, then focus on that.

Small-form-factor (2.5") disks are available in the following capacities:

72GB, 146GB, 300GB, 450GB, 600GB, 900GB, 1200GB in enterprise disks (10k/15k) and 500GB and 1TB for (5400/7200 RPM) drives.

Large-form-factor (3.5") disks were/are available in

146GB, 300GB, 450GB, 600GB capacities for 10k/15k RPM enterprise disks

and

500GB, 1TB, 2TB, 3TB, 4TB, 6TB in nearline/midline bulk-storage media (7200 RPM)

---

e.g. Buying a 600GB SAS 3.5" 15k RPM enterprise disk would be a mistake today, as would purchasing a 1TB SATA 2.5" 7.2k RPM drive. Both of those are well outside of the sweet spot and ideal application for their respective form factors.

A note about HP's ProLiant servers: Large-form-factor 3.5" disks are NOT featured prominently in the product line. You may see LFF disks in product photos and marketing material, but all of the product SKUs that you'll likely see in distribution are going to be SFF. Only a couple of low-end models of the DL380 Gen9, for instance, are spec'd with 3.5" disks.

---

Solution 2

It is a cost/performance vs capacity question.

2.5" HDD, at the same RPM/rotational delay, have a performance advantage versus their taller brother by the virtue of the smaller platter area. This in turn permit lower seek time (because the head had to travel a physically shorter distance). At the same time, this means that total platter area (read: capacity) is at about 50% maximum relative to 3.5" disks.

For example, even modern capacity-oriented (10K RPM) 2.5 HDDs are limited to below 2 TB (Hitachi Ultrastar C10K1800 is 1.8 TB, but many other drivers are significantly smaller, at capacity between 900 GB - 1.2 TB). Performance (15K RPM) 2.5 HDDs are even smaller, at capacity below 1 TB. For all intent and purpose these 2.5 HDDs are challenged by 2.5 SSDs, with much faster performance and capacity exceeding the 1 TB mark (with some drive, as the Intel DC3700 / 750 series, hitting the same 1.8 TB max).

At the same time 3.5 HDDs, after stagnating for years at the 2 TB mark, are now available at capacity of up 6 TB (both 5.4K and 7.2K RPM) and even 8 TB (Hitachi He8 and Seagate Archive serie, even if the latter is not recommended in the general usage scenarios).

This leads to many vendors proposing "convertible" chassis, where a basic design can be ordered with 2.5" bays or 3.5" ones, with a ratio of (often) 2X more bays for the 2.5" version. Let's say that our server of choice can both have 24 2.5" bays or 12 3.5" ones. If you build for performance, at the maximum 1.8 TB capacity for 2.5" drive you can have about 43.2 TB total capacity. With more affordable 1.2 TB disks, you are at 28.8 TB. Doing the same math with 3.5" HDDs, with 8 TB models you are at 96 TB, and with more affordable 6 TB ones you are at 72 TB. As you can see, 3.5" HDDs are good for 2X/3X increased capacity/density, at the cost of slower performance.

This is the exact reason why cloud providers (which often don't care much about performance, but are all about capacity) are going with 6/8 TB 3.5" HDDs. On the other side, virtualization and database workloads strongly prefers high-speed, lower capacity 2.5" HDDs (but really excels with 2.5" SSDs).

Solution 3

Another consideration is power usage (which is often billed). While 2.5" use less power per drive than 3.5" drives, because they are not available at higher capacities they use more power per GB.

Comments

[TRiG]

It also raises the question of how to deal with this constant misuse of the phrase to beg the question.

[dbr]

Perhaps it's just a useful turn of phrase that doesn't need addressing?

[tedder42]

The Innovator's Dilemma is literally about this topic- physical disk sizes.

[ewwhite]

@tedder42 Can you explain?

[dbr]

So for you, it really is just done to the issue of larger disks? In general, if you don't need those, you don't need LFF? Taking a look at the ML150 G9 for example, both the entry and base models ship with an LFF cage. Only the "performance" model ships with an SFF cage (taking the pre-built SKUs here - not BTO). Again, this is a low-end server, but why do you suppose it's configured that way?

[ewwhite]

Because people looking for low-end servers are likely to buy cheap, larger disks (SATA, in particular). Enterprise applications typically require SSD or enterprise SAS drives of lower capacity, but higher speed. An ML150 is at the bottom of the HP product line, so potential buyers may not want the expense of higher-end disks.

[dbr]

I appreciate all of that, but in my experience, LFF drives don't seem to be much cheaper than SFF these days (for capacities where both form factors are available), so why would a low-end LFF box actually be much cheaper than an SFF one with the same disks (aside from the form factor)? That leaves just the high capacity issue, which to me doesn't seem like something that is weighted towards small companies. Can you see my confusion?

[ewwhite]

@dbr There's isn't very much overlap in capacities. Can you provide specifics of what you're looking at?

[dbr]

Sticking with the HP example, their current range of SAS drives overlaps quite a bit - there's both SFF and LFF available in almost all the capacities between 300GB and 1.2TB (2TB if you include 512e options). Why would I ever want to choose the LFF options of the <1.2TB drives? Are they really only for people who also want to mix multi-TB drives in the same cage or perhaps people who already have a box with just LFF bays and need some more disks? If I'm spec'ing even a new entry-level box that doesn't need >1.2TB disks, I don't see much point in considering LFF. Does that sound about right?