You just activated my battlecard: How IBM sales droids plan to whack flash array rivals
Big Blue, what's wrong with Dell EMC, HPE, NetApp or Pure?
IBM's Evaluator Group has knocked up sales briefing battlecards that aim to kill off competition from Dell EMC, HPE, NetApp and Pure for its new FlashSystem 9100 flash arrays.
We've seen them and summarised the main points to be used by Big Blue and its channel sellers when coming up against these foes. We've also asked the aforementioned four vendors what they think and when they will add their commentary.
Dell EMC PowerMax
The PowerMax was thought to be a faster array than the FS9100 and IBM's battlecard noted this as a point to be defensive about. But the FS9100 actually runs up to 10 million IOPS (with cache hits) in its 4-node cluster config so the two are broadly level in performance. The other areas where IBM felt it came out on top included:
- Simpler to use through its GUI
- Ahead with NVMe over Fabrics plans where Dell EMC is limited to 10GbitE at this point and has made no decision on iWARP or RoCE, has not disclosed intentions
- Storage Insights better than analytics and reporting features on PowerMax, with CloudIQ not available
- Able to virtualize third-party arrays, which PowerMax cannot
- PowerMax has different code from other Dell EMC arrays with FS9100's Spectrum Virtualize software shared with SVC, V7000 and V5000
- Integrated container support with Spectrum Connect vs PowerMax's software plug-in
- FS9100 users can pay monthly for capacity used after data reduction while PowerMax has no comparable base offering
PowerMax has unified block and file support, with the FS9100 being block-only. IBM said combined block and file functioning on PowerMax would not be optimal for performance.
We note that, although Spectrum Virtualize code is shared with the SVC, V7000 and V5000, the FlashSystem 900, A9000 and A9000R run different code from the FS9100.
Compared to HPE's 3PAR array, IBM claims the FS9100 is:
- Faster with its NVMe drives. FS9100 has <100μs read times vs 3PAR's 500μs
- Ahead with NVMe-over-Fabrics plans where HPE has not disclosed intentions
- Simpler to use through its GUI
- Better instrumented for analytics
- Able to virtualize third-party arrays, which 3PAR cannot
- Better secured with encrypting Flash Modules and external attached storage vs 3PAR's restriction to self-encrypting drives
- Integrated container support with Spectrum Connect vs 3PAR's software plug-in
- Able to tier to the public cloud while 3PAR can not
- Better warranty at seven years vs 3PAR's five years
The battlecard suggests IBM sales people and partners could ask if 3PAR is suffering from lack of investment.
Once more, the 3PAR system has unified block and file support. Again, IBM claims combined block and file functioning would not be optimal for performance.
The lack of NAS support in the FS9100 is seen as an attackable weakness by IBM. It suggested customers should be asked if they really do need to mix block and file access, and affect system performance.
NetApp's recent A800 is attacked on several fronts:
- Clustering is complex to set up vs FS9100's simpler GUI
- Software-only compression vs FS9100's hardware assist
- Inferior ActiveIQ analytics compared to Storage Insights
- Block storage must be translated to files on a file system, meaning performance overhead
- Virtualizing external storage requires FlexArray feature on FAS; not available on A890 whereas FS9100 can virtualize external arrays
- FS9100 users can pay monthly for capacity used after data reduction while NetApp has no comparable base offering
- Integrated container support with Spectrum Connect vs NetApp's software plug-in
A NetApp strength is seen as its NAS support which the FS9100 lacks. As with HPE and 3PAR, the battlecard suggested customers should be asked if they really must mix block and file access, and affect system performance.
Pure Storage FlashArray //X
The recently updated FlashArray //X is the focus of a fourth battlecard.
- FS9100 is simpler to use through its better GUI
- No scale-out and 250μs response time vs FS9100's <100μs
- Faster, hardware-assisted compression in FS9100
- FS9100 can scale out capacity and performance where Pure scales up capacity, increasing performance by moving to a different model
- FS9100 shares Spectrum Virtualize software with SVC, V7000 and V500, whereas Pure has disjointed products
- FS9100 can virtualize third-party arrays, which Pure Storage cannot
- Integrated container support with Spectrum Connect vs Pure Storage's Flocker plug-in
IBM noted that Pure Storage has a good NVMe-oF story. It's committed to FC and Ethernet with RoCE, and supports 32Gbit/s FC and 10/256/40 GbitE.
The FS9100 is "committed to FC and Ethernet with iWARP and RoCE. It has 16Gbit/s FC and 25GbitE adapters. Defend that these are the most common adapter interfaces."
As with Dell EMC we point out that, although the FS9100's Spectrum Virtualize code is shared with the SVC, V7000 and V5000, the FlashSystem 900, A9000 and A9000R run different code from the FS9100. In other words, IBM could be said to have a disjointed product set.
Overall IBM identifies the FS9100's lack of file access support as a weakness against Dell EMC, HPE and NetApp. Against Pure its main identified threat is NVMe-over-Fabrics support. ®