While Exablox describes its object-based OneBlox appliance as a production storage system, many of its customers use it for backup. This week the vendor added a useful backup feature – variable-length deduplication.
Variable-length deduplication is an alternative to fixed-length deduplication. Variable-length dedupe breaks a file system into chunks of various sizes while fixed-length breaks all files into chunks that are the same size. Because it can use smaller chunks, variable length dedupe can get better reduction ratios.
Exablox senior director of products Sean Derrington said he expects a backup on OneBlox with variable-length dedupe to typically provide a 10:1 ratio compared to 3:1 for fixed length.
EMC’s Data Domain and Avamar, and Quantum DXi disk backup systems also use variable block dedupe.
Exablox storage is object based, but users access data through an NFS or SMB file share. All of its dedupe and compression occur inline, and it also supports continuous snapshots and replication between boxes.
Exablox supported fixed-length dedupe on OneBlox from the start. Derrington said he expects some customers will still use fixed-length for primary storage but variable-block will be the more popular option for backup. Customers can use fixed- length and variable-length for different applications inside the same storage pool.
“For any applications that they’re storing on OneBlox, customers can define storage policies on a per share basis or application basis,” Derrington said. “They can decide if they want fixed-length or variable-length dedupe, if they want compression on or off, snapshots on or off, or remote replication on or off. They can use fixed-length dedupe because it’s better suited for primary data and turn compression off on videos or images because they don’t compress well. They can turn snapshots off if the application does [snapshots].”
Derrington said a customer using fixed-length dedupe now can switch to variable-length, and get the full benefit of the better ratios after the current retention period passes. “If they have a 14-day retention period, on Day 15 all the data that’s been backed is on variable-length dedupe,” he said.
Tim Stammers, a senior analyst at 451 Research, said Exablox offers “simple cheap and deep storage” with a twist. “It’s unusual to have native NFS and SMB on an object box,” he said. “Exablox supports existing apps and leaves you with object storage underneath.”
Exablox also added an on-premise option for managing OneBlox appliances. Private OneSystem proactively monitors and identifies potential problems. It is deployed as a virtual machine inside a customers’ data center. From the start, Exablox used a cloud-based OneSystem for storage management. Now customers can choose between on-prem and cloud management.
Nakivo this week upgraded its Backup and Replication software, adding support for Synology RackStation and DiskStation NAS devices for virtual machine backup in VMware environments and to the Amazon Cloud.
In October, the company created a similar installer for Western Digital NAS devices for virtual machine appliances that are onsite and offsite. The latest software install supports up to 20 Synology NAS models and next year Nakivo plans to support QNap NAS systems.
“Our requirements are quite modest,” said Sergei Serdyuk, Nakivo’s director of product management. “We support everything that has one gigabyte of RAM and two CPUs. Our software runs right on the [NAS] box. You don’t have to go through any protocols like CIFS. The installment has been made very simple and should not be a problem.”
Nakivo Backup and Replication runs on a physical or virtual machine within a VMware environment and helps boost backup speeds when the software is deployed directly on a Synology NAS. The backup data is written directly to the NAS disks, bypassing NFS and CIFS protocols.
The backup and replication software can use the available space on the NAS device to store virtual machine backups. The VMware data is automatically deduplicated at the block level so that only the unique data is written to the virtual machine backup repository.
The data deduplication works on a global level across the entire backup repository so that all the data from all the virtual machine backups are factored in. After the virtual machine data is deduplicated, the software automatically compresses each block of data to save space in the backup repository.
The VMstore T5040 is the third model in the T5000 series launched in August. The T5000 series now matches Tintri’s flagship T800 hybrid platform with three versions, each tuned for a certain number of virtual machines.
The three T5000 models use the same dual-controller 2u box, each with 24 solid-state drives. The difference is in the capacity of the SSDs. The T5040 uses 240 TB SSDs and is rated for up to 1,500 VMs. The T5060 holds 480 TB SSDs for 11.5 TB of raw capacity and 2,500 VMs, and the T45080 uses 960 TB SSDs for 23 TB of capacity and 5,000 VMs. All T5000 systems ship fully populated with 24 SSDs.
“We had designed all three from the start, but wanted to test our customer base and see if there was demand for something lower than the 5060,” said Chuck Dubuque, Tintri senior director of product marketing. “A lot of our customer needs are still provided by hybrid systems, but we did have requirements for 100 percent flash for certain workloads.”
Dubuque said Tintri expected about 10% to 20% of its new systems sold would be all-flash when it launched the T5000, and that forecast has been accurate. He said he expects most all-flash arrays to go to T800 customers who want better performance for workloads that prove difficult to virtualize.
Tintri on Thursday will preview a new version of Tintri Analytics in an online presentation.
Dubuque said new predictive analytics features expected in 2016 will build on Tintri’s current real-time monitoring and troubleshooting. The new analytics will predict problems and the results of changes to applications. Tintri analytics are done at the VM-level.
“We have this rich data set from every single VM in terms of metadata about performance, size, how much flash it is utilizing on hybrid arrays, the name of the VM, which hypervisor it’s on,” he said. “Customers will be able to use that to predict future growth requirements much more accurately. We will use customers’ historical VM level data to model workload growth needs. For example, it will show if a customer is running low on flash and might run out of flash performance before running out of physical capacity. This will tell you the best solution – whether you should buy another hybrid system, an all-flash system, and what you need to do to rebalance your system if your needs change.”
Enterprise storage revenue increased 2.8 percent year-over-year in the third quarter of 2015 while networked storage declined 3.1 percent, according to IDC’s disk tracker numbers. Perhaps a better way to put it is Hewlett Packard Enterprise (HPE) and smaller vendors increased, while the rest of the large vendors’ sales tanked.
HPE’s overall disk sales increased 16% to $1.49 billion and its market share grew from 14.4% a year ago to 16.3%. HPE remained second but closed in on leader EMC, which dropped from 20.5% share to 18.4%. EMC’s disk revenue dropped eight percent over last year to $1.68 billion.
IDC said worldwide disk revenue totaled $9.1 billion for the quarter, and capacity shipments grew 31.5% to 33.1 exabytes. IDC credited a rise in server-based storage and storage sold to hyperscale data centers for the overall gain.
HPE was followed by No. 3 Dell with 9.9 percent market share and $899.4 million, No. 4 NetApp with 7.1 percent share and $651 million, and IBM with 6.4 percent share and $584.6 million. All three declined in revenue from last year. IBM’s 32.5% drop reflects it selling its server business to Lenovo.
Revenue from original design manufacturers (ODM) that sell directly to hyperscale data centers increased 23.4% to $1.25 billion and now makes up 13.7% of the market. The smaller vendors who make up the “others” category increased 15.2% to $2.58 billion and 28.2% share.
EMC still has a tighter grip on the external storage market, which consists of storage area networks and network attached storage. All of EMC’s $1.67 billion revenue was in this category, but it fell eight percent from last year and its share dipped from 30.7% to 29.1%. NetApp remained No. 2 despite a 12.8% drop to $651 million and 11.3% share. HPE increased 5.3% to $598.5 million and 10.4% to move past IBM, which stood fourth at $534 million and 9.3 percent market share after a 9.6 percent decline. Hitachi Data Systems rounded out the top five after falling 11.3% to $451.2 million and 7.8 percent share. All other vendors combined increased 8.4 percent to $1.84 billion and 32% share.
The overall networked storage market fell to $5.76 billion in the third quarter. IDC said all-flash array revenue increased 60.8% to $626 million in the quarter and hybrid flash array storage jumped 26.5% to $2.4 billion.
“Spending on traditional external arrays declined during the quarter as infrastructure refresh, coupled with the demand for software-defined storage and cloud-based storage, drove investments more heavily in server-based storage and hyperscale infrastructure,” IDC storage research manager Liz Conner said in the IDC press release.
Pure Storage’s first earnings call since its October IPO was a rosy one on the revenue side.
The all-flash array vendor’s revenue hit $131.4 million in its fiscal 2016 third quarter – a 167% increase over the $49.2 million for the same three-month period ending Oct. 31 a year ago. Pure Storage claimed to have added 250 new customers, including Domino’s Pizza and The Boston Globe, in the quarter to boost the total number north of 1,350.
Pure’s strong revenue growth stood out especially in comparison to the dismal earnings report Tuesday from another all-flash array (AFA) specialist. Violin Memory reported revenue of just $12.5 million – down 18% from the prior quarter and 42% over the same period a year ago.
Pure will rank third in revenue among AFA vendors, behind EMC and IBM, in an International Data Corp. (IDC) forecast due out next week, according to Eric Burgener, a research director in IDC’s storage practice.
Although Pure stands tall over other AFA startups, the company faces increasingly stiff competition from long-established storage vendors. EMC recently said its XtremIO AFA line saw triple-digit revenue growth year-over-year for the last two quarters and remains on track to do $1 billion in business in calendar 2015. IBM reported flash revenue increases of more than 50% in each of the last three quarters.
Among the other large vendors doing well on the flash front is Hewlett Packard Enterprise (HPE). CEO Meg Whitman said last week that revenue for HPE’s 3PAR AFAs more than doubled during the last quarter, and the product is on pace for $500 million in sales in the next year.
“Guys like EMC, IBM and HPE all have secondary storage platforms, and they can replicate between their primary all-flash arrays and any of their other platforms if they want to use them for backup, for disaster recovery, to move data into an archive,” Burgener said. “That’s going to be a challenge for companies like Pure that are currently selling only a primary storage array because obviously the other guys are pitching the full portfolio.”
Another challenge for Pure will be keeping expenses down to keep its losses in check. The company made progress last quarter, reducing its non-GAAP operating loss to $28.1 million compared to $51.6 million in the previous quarter.
Stifel Financial Corp. analyst Aaron Rakers wrote in a research note that he was more impressed by Pure’s operating leverage than by its revenue exceeding expectations by 16.2%. Stifel noted that Pure reported a decrease in non-GAAP operating expenses to $109.2 million – 5.4% lower than the financial firms’s expectations – although Pure did push out some marketing and engineering spend to this quarter.
In a blog post, Pure CEO Scott Dietzen wrote that the company added $1.70 in revenue for each additional $1 of operational expense. Dietzen claimed Pure would continue to improve its operating efficiency as it invests aggressively in sales, marketing, support and its channel to maximize growth.
“If you look at Q3 this year vs. Q3 last year, we’ve cut our operating losses more than in half,” said Pure CFO Tim Riitters, citing the shrinking operating losses as a reflection of the company’s “discipline and focus.”
Riitters said Pure expects to generate positive cash flow in 2018.
“Right now we’re still only about a ,1200-person organization worldwide. EMC is a much, much larger organization,” Riitters said. “One of the reasons we’re growing and investing so much is really to increase the number of what we call at-bats. We’re constrained by the ability just to get in front of customers and show them how profoundly different the Pure Storage technology is. When we do that, we win our fair share.”
All-flash pioneer Violin Memory can’t sell enough all-flash arrays to make money. So its next move could be to sell itself in its entirety.
After another disappointing quarter, Violin CEO Kevin DeNuccio said the company has hired an investment banker to explore “strategic alternatives.” That usually means the company is for sale. But are there any takers with a glut of flash arrays on the market?
“There is no set timetable for completing the process, nor are there any assurances given that the exploration of strategic alternatives will result in any transaction being consummated,” DeNuccio said on the earnings call last night.
A financial analyst on the call asked DeNuccio what type of company might be interested in Violin.
“We think this is an attractive asset, both from a partnership or acquisition potential … “ he said. “The industry is obviously restructuring pretty dramatically with several companies buying storage companies and this company buying flash fabs. So we think there is a broad range of people interested in an asset like this.”
Violin hasn’t been able to find enough people interested in buying its flash arrays. Its $12.5 million in revenue was down 18% from the previous quarter and 42% from last year, and below Violin’s forecast of $16 million to $20 million. Violin lost $22.7 million in the quarter compared to $24.4 million the previous quarter and $23.5 million in the same quarter last year.
Perhaps the most disappointing aspect of last quarter was product revenue fell 63% to $6.3 million in the second quarter of availability of Violin’s Flash Storage Platform (FSP). Violin bet its business on its FSP, which added storage management and data protection features that were missing in its earlier performance-focused arrays. The FSP is still in early days and Violin added two new models this week, but a paltry $6.3 million in sales in a growing market doesn’t give much hope. Pure Storage, another all-flash pioneer, reported $131.4 million in total revenue and $113.6 million in product revenue in the same quarter.
DeNuccio admitted the quarter results were “extremely frustrating,” He said Violin failed to close one multi-million dollar deal involving a customer of its older 6000 technology that pushed back a FSP 7000 Series rollout.
“Our FSP and Concerto software, which consisted more than 50 million lines of code, has taken significant time and rigorous life testing to address the nuances of data risk, customer environment and to get things fully stabilized and operating properly,” he said.
He said Violin did gain a large repeat order of an FSP array from a large U.S. cable company and expects to add more than $1 million in revenue from that company each quarter. “This win … affirms our strategic premise,” Nuncio said.
Now if that cable company wants to get into the flash storage business, maybe Violin can find a buyer.
HP CEO Meg Whitman made that clear Tuesday during HP’s last earnings call before the split. Whitman, who will lead HPE, gushed over the momentum of 3PAR ahead of the company break-up.
Whitman said 3PAR StoreServ picked up share last quarter in the overall storage and flash-only markets. She said 3PAR all-flash arrays are on pace to sell $500 million over the next year and revenue more than doubled last quarter. That puts it ahead of the growth of newly public all-flash specialist Pure Storage, although it is still playing catch up to EMC’s XtremIO market leading all-flash array. Whitman also pointed to a recently released Gartner report that ranked 3PAR first in critical capabilities for consolidation, OLTP, virtualization/VDI, analytics and cloud use cases as well as first among high-end arrays.
Overall, HP storage revenue declined seven percent last quarter but Whitman claimed “storage significantly outpaced the market.” HP executives claim they picked up 10 percentage points of share in the all-flash market in the past year.
“Obviously the storage business is doing very well with 3PAR and all-flash,” said Whitman, who added that HPE will become more aggressive to take advantage of confusion created by the Dell-EMC merger.
If the largest storage array vendors that sell Brocade’s SAN switching are getting hit with declining revenue every quarter, that’s bad news for Brocade, right?
If you agree with that, then you’re missing the big picture, according to Brocade CEO Lloyd Carney. Carney said it doesn’t matter if people are buying their storage from big vendors or small, using flash or hard disk drives, or putting their data on premise or in the cloud. He says that if storage capacity grows, that’s good for Brocade because Brocade is one of only two vendors that sells the switching – Fibre Channel and Ethernet – that storage systems require. And storage capacity is growing, even if people are paying less for it.
“The knee-jerk reaction that a lot of us have is, storage dollars are coming down so something must be wrong with Brocade,” Carney said on Monday’s Brocade earnings call with analysts. “Well as long as storage capacity grows, Brocade is fine.”
At another point on the call, he said: “As long as storage capacity is growing, whether it’s Fibre Channel or IP-based, we will be successful.”
Carney’s optimistic outlook was justified last quarter. Brocade’s $588.8 million in revenue was better than its previous forecast, and its $324.9 million of SAN revenue was far above expectations. Overall SAN revenue was about the same as last year while most storage vendors saw revenue declines last quarter, and Brocade’s large director switch revenue improved 14% over last year.
Carney was less optimistic for this quarter, though. Brocade’s forecast of revenue from $550 million to $570 million is below the $589 million it did in the same quarter last year, and it expects SAN revenue to grow no more than three percent compared to an average gain of nine percent in this quarter. Brocade still predicts growth for the full year in 2016, however.
Carney said Brocade won’t get squeezed like the big storage vendors because it doesn’t face nearly as much competition. There are a bunch of smaller storage array vendors trying to take customers from the likes of EMC, NetApp, Hewlett Packard Enterprise, Hitachi Data Systems, IBM and Dell. On the SAN switching side, it’s Brocade and Cisco. And SAN switching is a small part of Cisco’s business.
“We are better positioned than the people who sell raw storage from a competitive standpoint,” he said.
Carney said Brocade’s plan to beat Cisco is to innovate – he pointed to non-disruptive monitoring and analytics added last quarter – and expand its partner base. Brocade recently broadened partnerships with Huawei and Lenovo with the intention of expanding its business in China.
As long as storage capacity expands, Brocade’s CEO expects to take in its fair share.
“Despite what is happening at EMC or NetApp or overall, people are buying more storage,” Carney said. “You yourself are using more storage. You’re sending pictures, you’re sending video content. So everybody is using more storage.
“Now, the price per megabyte or per terabyte of storage is coming down, because it’s such a competitive marketplace. However, the connectivity into that storage, the layer at which we play, is not as competitive a place.”
Nimble Storage’s drive for profitability took a major hit last quarter. Nimble, which was believed to be one of the smaller competitors eating away at large storage vendors’ sales, won’t hit break-even this quarter as it previously forecasted.
Nimble’s revenue of $80.7 million was up 37% from last year, but roughly the same as the previous quarter and below the hybrid array vendor’s forecast of from $86 million to $88 million. Nimble lost $11 million in the quarter – its largest quarterly loss since going public in 2013 — and increased spending this quarter will push back profitability indefinitely. Nimble forecasted revenue of $87 million to $90 million and losses in the $8 million to $10 million range for this year-ending quarter, typically the best sales period for storage vendors.
Reaction from Wall Street was potentially devastating. Nimble’s share price fell from $20.39 to $10.00 overnight – a drop of nearly 50% — and at least 11 financial analyst firms downgraded the stock.
CEO Suresh Vasudevan blamed the problems on two things. He said Nimble is affected more by large vendors’ price cuts as it moves deeper into the enterprise. And it has struggled to balance growth in the enterprise and its traditional commercial markets while trying to control spending with an eye on profitability. He said Nimble will beef up its commercial sales staff and is working on an all-flash array to help enterprise sales.
“We now believe that this approach of constraining investments at the same time that we diversify our customer base may have impacted our growth,” Vasudevan said on Nimble’s earnings call Thursday night.
He repeated what he has hinted at in the past – that Nimble will add all-flash arrays. Nimble has an all-flash expansion shelf and its arrays can pin data to all-flash volumes, but it lacks an all-flash product as more enterprises are looking to go in that direction.
“We have said that we have a very concrete plan for broadening our flash platform to compete in the entire space and that is still very much on target,” Vesudevan said. “You should expect to see us participating in the entire market with both hybrid flash and all-flash.”
The price cuts from the likes of EMC, NetApp and Hewlett Packard Enterprise could be a tougher problem to solve. Vasudevan really had no specific counters when asked his plans for dealing with that.
“We believe our business foundation remains extremely strong,” he said.
That belief will be tested in coming months. Meanwhile, the drastic drop in Nimble’s stock price is likely to attract some potential suitors who want to broaden their storage portfolios in the wake of the Dell-EMC deal.
NetApp’s latest earnings report told a familiar story.
NetApp Wednesday night said its revenue last quarter – although higher than expected – continued to decline with the vendor’s most established products taking the biggest hits. Its newer flash, cloud and software-defined software products are on the upswing but not nearly enough to keep overall sales up.
These same trends have been hitting NetApp and other large storage vendors for more than a year, and led to EMC’s getting swallowed whole by Dell. They contributed to NetApp switching CEOs from Tom Georgens to George Kurian in June without much change in result. And the trends will almost certainly continue for the foreseeable future.
NetApp’s revenue of $1.45 billion last quarter declined 6.3 percent from last year, and product revenue fell 12% to $819 million. For this quarter, NetApp expects revenue of $1.4 billion and $1.5 billion, or roughly a seven percent decrease from last year.
“Parts of our business are working well, some parts need improvement and other parts we must manage through declines,” Kurian said on the earnings call. “The IT spending environment continues to be constrained and the expectation for growth for the overall storage market has decreased to low single-digits.”
The parts that are working well, he said, are scale-out, software-defined, flash, converged and hybrid cloud storage. Those areas are growing about 20% a year while the traditional standalone hybrid storage market declines approximately nine percent, he said.
For NetApp, that means its Clustered Data OnTap (CDOT) operating system, hybrid cloud software, and all-flash arrays are growing while its traditional FAS with Data OnTap 7-Mode and its OEM products decline.
Kurian said 7-Mode OnTap shipped on about 30% of NetApp FAS arrays last quarter, down from around 65% a year ago. CDOT was on nearly 70% of new FAS systems shipped in the quarter, up from 35% last year. CDOT systems grew 95% year-over-year while 7-Mode shipments declined 60%. Still, CDOT overall made up only 17% of NetApp’s revenue compared to 15% in the previous quarter. That means 7-Mode customers are still not upgrading in large numbers.
NetApp’s All-Flash FAS array unit shipments increased 445% year-over-year following a price cut, controller upgrade program and seven-year extended warranty in June.
Kurian said Dell’s planned $67 billion acquisition of EMC is “clearly an opportunity” for NetApp because it will create confusion for customers and the channel. “The Dell-EMC transaction is yesterday’s solution to tomorrow’s customer problems,” he said. “It does not fundamentally address the hybrid cloud, it does fundamentally address the data management opportunity that customers are forced to deal with. It is really about trying to build efficiency in an integrated hardware business rather than the software-defined data center of the future.”
NetApp has $4.8 billion in cash, but Kurian didn’t sound enthusiastic about making acquisitions to bolster its product portfolio. When asked if he would consider that, he said NetApp’s strategy focused more on growing the emerging products it already has.