The Future of Storage The disruptive transition to software-defined Sizing the opportunity November 3 rd, 2014 By Jérôme Lecat, CEO of Scality
I. Executive Summary Software is eating the world Marc Andreessen famously stated, Software is eating the world. 1 Beyond the strong prophecy, it is interesting to look at what it really means for the IT industry, and specifically, for the $100 billion storage industry. Many of us have gone through several technology waves, and the pattern repeats itself time after time, in what is known as the S-curve. 2 When a new technology disruption is introduced, the observers are divided between the bullish, who are overly optimistic about the speed of change, and the doubters, who believe that the new technology will never replace the status quo. Indeed, it is extremely difficult to state at the introduction of a new technology whether it will succeed, but when it does, its start is slow, but ultimately its supremacy is often beyond the bullish expectations. At Scality, we believe that software will eat the storage industry; we call it Software Defined Storage (SDS). The disruption that will happen to the IT storage industry is similar to the disruption brought onto telecom by the Internet or to retail by Amazon: a complete change in the value chain. As a result, traditional vendors will lose billions of dollars in business while new ways of doing things will attract billions of dollars in business opportunity. Software running on commodity hardware is the prime enabler of this disruption. Scality embodies this transformation, with field-proven software that operates on any commodity server, without any certification program or other hardware-attached attributes. Scality s software, the RING, has been designed from scratch to bring to large enterprises the solutions used by Internet giants. This study aims to quantify a realistic estimate of what the disruption looks like in the next 6 years, with a 2020 horizon. We will especially focus on scoping the addressable market for Scality: the software opportunity in capacity-driven Software Defined Storage. By 2020, the storage industry will be dramatically different: Less than 25 percent of storage spending will be in small and medium size datacenters Ten years from now, a limited number of small enterprises will operate their own IT. They will leverage cloud and SaaS services such as Salesforce, Office 365, Box or Amazon s AWS. In its Datacenter Forecast, 3 Gartner reports that in 2010, 42 percent of datacenter spending on x86 servers and storage systems was for datacenters with fewer than 100 racks, and by 2018, this number will have fallen to less than 30 percent. We expect this trend to continue so that by 2020, less than 25 percent of storage spending will be for smaller datacenters. In essence, 17 percent of the market will disappear for traditional vendors. Obviously, there will be a storage opportunity for the cloud and SaaS providers, but it will be of a different nature. 40 percent of storage capacity will be sourced directly from disk and component manufacturers Historically, established storage vendors like EMC or NetApp have enjoyed gross margins of about 55 percent; an impressive level when compared to other players. Hard disk drive (HDD) manufacturers (Seagate, Western Digital) typically achieve 30 percent margins, while original equipment manufacturer (OEM) server brands (like HP and Dell) are below 30 percent, and Taiwanese server manufacturers (like Quanta) are below 10 percent. Much of EMC and NetApp margins come from the high-price resale of components, including HDDs. The availability of Software Defined Storage will put tremendous pressure on such margins, as it becomes easy for large buyers to procure the software and commodity server separately. Furthermore, since reliability is built in the software, not the hardware, customers are incented 1 Why Software Is Eating The World by Marc Andreessen, http://online.wsj.com/articles/sb10001424053111903480904576512250915629460 2 http://en.wikipedia.org/wiki/technology_life_cycle#s-curve 3 Gartner: Forecast: Data Centers, Worldwide, 2010 2018, published August 2014 G269529 2 Scality 2014
to buy lower quality servers. As a consequence, we expect original design manufacturers (ODMs), both server ODM and HDD manufacturers, to sell systems directly to the larger datacenters. Already today, some large public cloud providers and other large storage users procure their HDDs directly from the manufacturers (mostly Seagate and Western Digital/HGST). According to IDC, this business grew from a mere $0.4 billion in 2010 to $2.1 billion in 2014, and will grow to $3.1 billion by 2017. We believe that this phenomenon will accelerate, enabled by the emergence of SDS software. Investment bank J.P. Morgan estimates that 30 to 35 percent of the server market will shift to ODM in the next 5 years. 4 Most large enterprise and service providers will continue to operate their own IT, but will do so with a software-led approach While small and medium size businesses will leverage cloud services; the service providers, large enterprises and many vertically focused players will continue operating their own IT, but they will do so in a way similar to Google, Amazon and Facebook. This group represents about 20,000 entities worldwide that will be leveraging standard servers with software providing both the application and the infrastructure intelligence. By 2020, the transition will have already begun, and about 50 percent of these large enterprise and service providers will already have transformed most of their IT operations. Scality addresses 80 percent of storage workloads, representing one of two storage tiers in the softwareled future: Analysts often assess Scality as part of the object storage market, which is usually defined as both the internal organization of data on HDDs and as the interface to the application. Defined as such, the object storage market is seeing moderate growth, because only new applications (e.g. mobile applications and websites) have adopted object interfaces to date. Most enterprise applications rely on traditional file and block interface methods. The Scality RING is architected around an object storage core technology, but it presents traditional interfaces to applications. As a consequence Scality regularly competes with or replaces traditional SAN, NAS and scale-out NAS systems. Indeed over 50 percent of Scality s customers use RING as a scale-out NAS. Scality addresses a market that is much wider than analyst-defined object storage market. We believe that in the software-led world, the storage infrastructure will be collapsed to two tiers. For applications that are latency sensitive, like ERP, relational databases, VDI, or high-frequency trading, the storage will be provided in silicon. Several architectures are competing, including all-flash arrays like Pure Storage, VM-optimized hybrid-storage like Tintri or converged systems like Nutanix. For all other applications, including SaaS, user generated content, documents, content and video streaming, large data repository, big data, distributed computing and long term active archive what we call capacity-driven storage the preferred method of storage will be a scale-out, object-based software, like Scality s RING, deployed on commodity servers. The idea of storage moving to a two-tier architecture, and latency (not IOPS) being the key differentiator, is not new. As a point of reference, Richard McDougall, from the office of the CTO of VMware, explores this idea in an article titled, Is Cloud Storage going to disrupt Traditional Storage? Part 1: The demise of expensive datacenter storage 5 in 2010! Cloud storage is only now starting to be a reality profoundly affecting market dynamics. 4 https://www.ventureoutsource.com/contract-manufacturing/focus-odm-quanta-it-shift-cloud-infrastructure-leaving-dell-hp-traditional 5 http://blogs.vmware.com/cto/is-cloud-storage-going-to-disrupt-traditional-storage-part1-the-demise-of-expensive-datacenterstorage/ 3 Scality 2014
Altogether, tens of billions of dollars of value will shift in the new storage landscape: The storage industry as a whole will not grow beyond $100 billion, as new cheaper storage replaces more expensive traditional options The traditional storage systems market will shrink from $38 billion to $20 billion in six years (10 percent negative CAGR) The opportunity for software defined capacity-driven storage will be $42 billion, of which the market opportunity for Scality 6 is $5 billion This represents a huge change for the IT industry and its vendors. The mid-range of the market is essentially going to disappear, while the high-end will be completely transformed. IDC talks about the IT industry transitioning to the 3 rd platform, while Gartner calls this transformation the Web-scale singularity and states that during the past decade, leading cloud services organizations have been experimenting with new ways to deliver IT services. Their efforts have resulted in a singularity event that changed the trajectory of the IT landscape. What is being left behind is IT conventional wisdom 7 Wikibon, an analyst firm specialized on the storage market, calls this transition the Server SAN and writes Server SAN is software-led storage built on commodity servers with directly attached storage (DAS). Wikibon believes Server SAN is poised to disrupt traditional storage architectures over the next decade. As enterprise organizations begin to replicate the infrastructure of hyperscale giants, software will lead the transition. 8 A recent article from The Register titled SDI (Software Defined Infrastructure) wars: EMC must FORGET ARRAYS, adapt or disappear 9 discusses the impact of such drastic change on the established market leader. When asked how big is the market opportunity for the software that will be at the heart of future storage infrastructure, ESG founder, Steve Duplessie responded: You would need a crystal ball to answer this question. It is going to be a completely new order, but I d say that with a $5 billion estimate you are very conservative. The body of this paper will present the four detailed approaches we used to size the disruption of the storage industry: 1. Market Size per IDC File and Object Based Storage Forecast 2. Market Size per Scality s Analysis of Industry Trending 3. Bottom-up Approach from Vendor Revenues 4. Bottom-up Approach from the Number of Potential Customers 6 Total Addressable Market (TAM) 7 Gartner: Strategic Technology Trend: Web-Scale Singularity means goodbye to conventional IT wisdom. G00259636 8 Wikibon http://wikibon.org/wiki/v/the_rise_of_server_san 9 http://www.theregister.co.uk/2014/10/20/sdi_wars_emc_must_forget_arrays_adapt_disappear/ 4 Scality 2014
II. Market Size per IDC File and Object Based Storage Forecast IDC is the first analyst firm to have identified that the traditional segmentation of SAN, NAS and Object Storage is becoming irrelevant for the buyers. What matters are the workloads that the storage can fulfill, and whether the design is scale-up or scale-out. IDC has been stating that the bulk of storage growth will be coming from file storage for several years. By 2017, over 80 percent of storage capacity will be unstructured data, which will be served in the form of files or objects. 10 For this reason, IDC has merged the File Storage and the Object Storage categories to create a File and Object Based Storage (FOBS) forecast. 11 This is especially relevant to Scality because Scality competes as much with file storage vendors as with object storage vendors. This being said, IDC s FOBS forecast does not cover 100 percent of Scality s use case, because Scality can also deliver storage of virtual machines (similar to Amazon EBS) and Scality often competes against high-end and mid-range SAN (for consumer webmail storage for example), both of which are not covered in this study. The numbers below look at the growth of four market segments. These are segmented based on the internal organization of the content on the storage system, but all four categories present a file system to the applications. 1. File storage on a stand-alone server, which is a decreasing business in the coming years 2. Scale-up file-based storage, which corresponds mainly to traditional NAS and which is a decreasing business 3. Scale-out systems that have an internal file organization, like EMC Isilon and IBM GPFS. Scality fully competes with these solutions, and we believe that IDC has well represented the market adoption of object-based solutions over file-based solutions by their respective growth rates 4. Scale-out systems that have an internal object-based technology. This is the category that Scality wants to dominate. 10 IDC: Structured versus unstructured data: The balance of power continues to shift #247106 11 IDC: File and Object Based Storage Forecast May 2013 - #242287 5 Scality 2014
These numbers are in millions of dollars (USD), and include the underlying servers and storage media (flash and disk) for solutions that are delivered as software: Category Category example 2011 2012 2013 2014 2015 2016 2017 Unitary hostbased file Red Hat Enterprise Linux, SUSE Enterprise Linux, Microsoft Windows $5,500 $5,400 $5,300 $5,100 $4,900 $4,700 $4,600 server (scaleup) Server 2012, and proprietary Unix platforms such as Oracle's Solaris, HP's HP-UX, and IBM's AIX. Scale-up filebased software and appliances and gateways File-based software and appliances (scale-out) Object-based software and appliance (scale-out) and related gateways Windows Storage Server file servers from Dell and HP, ZFS-based appliances from Oracle and other suppliers, VNX NAS gateways from EMC, and NetApp FAS and V-series appliances running ONTAP in 7-Mode configurations, downloadable products from Nexenta, Tegile, and others. IBM SONAS (based on IBM GPFS), Quantum StorNext Appliances, EMC Isilon, HP StorageWorks X9000 Network Storage Systems, Dell FS Series, and NetApp's FAS appliances running Clustered ONTAP. This subsegment also includes a small percentage of disk-based scale-out file virtualization solutions such as those from Avere. Red Hat Storage Server (based on GlusterFS), Lustre-based variants, IBM GPFS, and Quantum StorNext software. CompuStorage solutions such as those from Nutanix, Pivot3, Scale Computing, and others are also included in this category. Servers for HDFS deployment are also part of this subsegment. EMC Atmos, NetApp Distributed Content Repository, Cleversafe dsnet, Amplidata's AmpliStor, Basho RiakCS, Cloudian, Exablox, DataDirect Networks Web Object Scaler (WOS), and Scality's RING. Examples of gateways are from suppliers like Nasuni, Riverbed, Panzura, TwinStrata, CTERA Networks, and Microsoft (StorSimple). $6,800 $6,700 $6,700 $6,700 $6,700 $6,600 $6,400 $2,000 $2,500 $2,800 $3,400 $3,900 $4,400 $5,300 $4,300 $6,500 $8,400 $11,400 $15,400 $18,100 $21,700 According to IDC, object-based systems will grow in market share from 23 percent of the revenue in 2011 to 57 percent exemplifying the transition described earlier. From a revenue perspective, it grows from $4.3 billion in 2011 to $21.7 billion by 2017, a 30 percent growth rate. Object-based systems are consumed by public clouds, private clouds and other uses (such as video distribution, active archive and distributed computing). Large public clouds, especially those of Amazon, Google and Facebook, have been able to bypass traditional hardware manufacturers and supply their storage from disk manufacturers. They typically develop their own object storage or leverage open-source software in a Do-It-Yourself approach. While they participate in the SDS market, they do not generate a potential market for software vendors. Other public providers buy hardware from server manufacturers, and choose to purchase software from SDS vendors like Scality. Large enterprises, such as banks will increasingly build private clouds, and they are likely to leverage commercial software to do so. Similarly, government agencies, distributed computing and specific verticals will deploy object storage to store vast amounts of unstructured data that are core to their operations. Given the value of the data for their activity, they are very likely to leverage commercial software. 6 Scality 2014
In our analysis, we break the object-based software and appliance number into five categories: 1. Public clouds purchasing HDD directly from ODM such as Seagate and Western Digital/HGST. 12 Typically, these players have enough engineering resources to develop their own software. 2. Public clouds purchasing from traditional vendors or deploying commercial SDS solutions like Scality 13 3. Private cloud 14 (by difference) 4. Other use of object-based storage (by difference) 5. Gateways, which is included in the scale-out number by IDC Scality then calculated the commercial software opportunity for each category: For public clouds, Scality assumes that those who Do-It-Yourself do not represent a commercial opportunity, and has made the very conservative assumptions that only 25 percent of the others will leverage a commercial software. Scality also assumes that the value of such software is only 10 percent of the total cost of the scale-out object-based storage. 15 The software opportunity is thus 2.5 percent of the storage number for the category. For enterprise private cloud and other use of object storage, it is assumed the proportion of those buying commercial software will rise from 50 percent in 2012 to that 75 percent in 2017, as enterprises that move to production workloads have a tendency to move away from open source and prefer an enterprise grade solution. Furthermore, it is assumed that the relative share of the software in the total cost will decrease from 30 percent in 2012 to 25 percent in 2017 as platforms grow. Object-based software and appliance (scale-out) Of which Public-cloud DIY - purchase from ODM Public-cloud - purchase from vendors Private Cloud Other use of object-based storage Gateways Commercial software SDS opportunity Public-cloud DIY - purchase from ODM Public-cloud - purchase from vendors Private Cloud Other use of object-based storage 2011 2012 2013 2014 2015 2016 2017 $4,300 $6,500 $8,400 $11,400 $15,400 $18,100 $21,700 $720 $1,680 $2,280 $2,310 $2,530 $2,970 $3,410 $1,374 $1,424 $1,804 $2,052 $2,334 $2,434 $2,650 $1,388 $1,686 $2,034 $2,514 $2,928 $3,276 $3,558 $718 $1,580 $2,113 $4,305 $7,324 $9,051 $11,603 $100 $130 $169 $219 $284 $369 $479 $- $- $- $- $- $- $- $34 $36 $45 $51 $58 $61 $66 $194 $253 $324 $422 $514 $596 $667 $100 $237 $337 $723 $1,285 $1,647 $2,176 Total TAM $328 $526 $707 $1,197 $1,858 $2,304 $2,909 Based on these assumptions, the file-and-object based storage commercial software opportunity grows at 34 percent a year, from 2014 to 2017, to reach $2.9 billion by 2017. The transition from conventional IT to web-scale IT will not be complete by 2017, and Scality expects the object-based system market to continue growing, at 25 percent per year, leading to a $42 billion dollar market, of which the software opportunity is $5.5 billion by 2020. 12 IDC identifies the storage spending of Public Cloud and Private Cloud in document #244924 13 IDC identifies the storage spending of Public Cloud and Private Cloud in document #244924 14 IDC identifies the storage spending of Public Cloud and Private Cloud in document #244924 15 Typically, the relative cost of software decreases with the size of the platform. The assumption is that public cloud deploy very large platforms. 7 Scality 2014
III. Market Sizing per Scality Analysis of Industry Trends IDC s Worldwide Enterprise Storage Forecast 16 presents the following numbers, showing the storage industry growing 5 percent a year in the coming years: Worldwide Enterprise Storage Systems Hardware, Software, and Services Customer Revenue, 2009-2017 (USD in billions) 2009 2010 2011 2012 2013 2014 2015 2016 2017 2013-2017 CAGR (%) Enterprise storage systems 26.0 30.7 33.6 35.1 34.6 35.7 37.2 38.8 40.3 3.9 Storage services 27.9 28.8 30.2 31.8 33.6 35.5 37.6 40.1 42.7 6.2 Storage software 11.6 12.7 13.9 14.1 14.8 15.5 16.3 17.1 17.8 4.8 Total 65.5 72.2 77.7 81.0 83.0 86.8 91.2 96.0 100.8 5.0 This forecast is partial, as it does not include cloud storage revenue (an alternative to on-premise storage). Neither does it include the spending in servers dedicated to storage by Public Cloud, nor the full cost of servers dedicated to storage in enterprises. We have used several other IDC documents, as well as consultancy with IDC, Gartner and ESG, and our analysis described in the previous section, to split the market as follows: 16 IDC: Worldwide and US Enterprise Storage Systems 2014 2018 Forecast May 2014 - #248554 8 Scality 2014
Source Scality from analyst data (see notes) 2014 1 2 Latencysensitive Capacitydriven In millions Storage Systems (SAN, NAS, Scale-Out NAS, All Flash Arrays sold as systems) - EMC, Netapp, HDS, HP, Dell, IBM, Fujitsu, Oracle, DDN 17 $ 10,647 $ 13,013 Storage Systems (SAN, NAS, Scale-Out NAS, All Flash Arrays, Converged Storage) - Other 18 $ 1,855 $ 2,268 3 Storage Software - EMC, IBM, HP, Netapp 19 $ 2,329 $ 5,433 4 Storage Software Other 20 $ 902 $ 2,105 5 Server Storage (of which HP 38% and Dell 25%) 21 $ 10,125 6 7 Storage purchased directly from ODM by Public Cloud and Private Cloud 22 $ 2,300 Storage Software - Scale Out Object Based Software Defined Platform 23 $ 1,197 8 Basic Cloud Storage (StaaS) - excludes SaaS like Box, Office 365 or Gmail 24 $ 1,500 9 Symantec Storage Software 25 $ 2,097 10 Tape Drive and Automation 26 $ 1,800 11 Storage Professional Services 27 $ 35,500 TOTAL $ 93,071 Lines 1 to 4 which total $38 billion represent what is generally referred to as traditional Storage Systems and Software. It includes SAN, NAS and Scale-Out NAS sold as an appliance, as well as related storage management software, and backup and archival software. Actually this number also includes some newer forms of storage like All-Flash Arrays and Converged Storage, but for 2014 the related number is small enough that it is insignificant. Note that while if we restrict our scope to Storage Systems as defined by IDC in its Worldwide Enterprise Forecast, Capacity storage represents roughly 72 percent of capacity for 55 percent of the value of the 17 IDC Worldwide and US Enterprise Storage Systems 2014 2018 #248554 complemented with Consultancy to separate key players from others The notion of Storage System corresponds as storage delivered in the form of hardware, ie appliance. In 2014, roughly 20 percent of the data is Latency-sensitive and represents 45 percent of the value of the Storage Systems and associated Storage Software. 18 IDC Worldwide and US Enterprise Storage Systems 2014 2018 #248554 complemented with Consultancy to separate key players from others 19 IDC Worldwide Storage Software 2013 2017 Forecast and 2012 vendor Shares #243435 20 IDC Worldwide Storage Software 2013 2017 Forecast and 2012 vendor Shares #243435 21 Referenced as Internal Storage in IDC Worldwide and US Enterprise Storage Systems 2014 2018 #248554 22 IDC Worldwide Enterprise Storage for Public and Private Cloud #244924 identifies the specific spending direct to ODM 23 Derived from IDC File and Object Based Storage Forecast as described in the previous section 24 Gartner Forecast: Public Cloud Servcies, Worldwide, 2011 2017 #248728 25 IDC Worldwide Storage Software 2013 2017 Forecast and 2012 vendor Shares #243435 26 IDC 27 IDC Worldwide and US Enterprise Storage Systems 2014 2018 #248554 and Worldwide Storage Service Forecast #244352 9 Scality 2014
storage industry. When we take a full view of the industry, including spending by public cloud, it is a much higher proportion. This section attempts to capture the trends of the storage industry for the next six years. The trend of each number of the previous table has been analyzed independently, and the consistency has been checked. Here are some of the key hypotheses that we have made: The study is based on the product set of each company as they are today (Oct 30, 2014). If there are cross-category buy-outs (which we expect), this will affect the table. We expect latency-sensitive storage to migrate from traditional storage systems to All-Flash Arrays and Converged Systems. More specifically, we expect the transition to be slow in 2015 and 2016 and accelerate in 2017. 28 We expect capacity-driven storage to migrate from traditional NAS and scale-out NAS towards software defined scale-out object-based architecture, as described in the previous section We expect the storage software sold by the traditional storage leaders, EMC, NetApp, IBM and HP, to be under strong pressure as their sales is very much linked to the traditional storage sales. We see their software revenue diminishing in the same proportion as the storage systems revenues. We expect other storage software to decline as well, as many of the software point solutions of today (such as WORM, Tiering, replication, retention policies and basic HSM) will be subsumed in the features of Software Defined Storage We expect more and more large public cloud and private cloud customers to procure systems directly from ODM, both from Asian hardware manufacturers and directly from the disk manufacturers for HDD. This transition will be made easier by the availability of Software Defined Storage. We expect the growth of cloud storage to be a new opportunity in 2015 and 2016 continuing to fuel the growth of the storage industry, but we expect that starting in 2017 it will start cannibalizing the growth of other forms of storage revenue. Actually, we expect service providers (like Box, Dropbox, Netflix or Smugmug), not end customers, to be the main use of Basic Cloud Storage, but these service providers will sell cloud services to end users, who in turn will not need to buy traditional storage. We have assumed that tape will remain stable. We expect professional services to increase by 4 percent a year in 2015 and 2016 (IDC estimate), but contrary to IDC, we expect storage professional services to stabilize and even decline after that, since Software Defined Storage requires much less human intervention than traditional storage. 28 10 Scality 2014
As a result of these assumptions, our estimate for 2020 is as follows: Source Scality from analyst data (see notes) 2014 2020 1 2 In millions Latencysensitive Capacitydriven Latencysensitive Capacitydriven Storage Systems (SAN, NAS, Scale-Out NAS, All Flash Arrays sold as systems) - EMC, NetApp, HDS, HP, Dell, IBM, Fujitsu, Oracle, DDN 29 $ 10,647 $ 13,013 $ 6,304 $ 4,810 Storage Systems (SAN, NAS, Scale-Out NAS, All Flash Arrays sold as systems, Converged Systems) - Other 30 $ 1,855 $ 2,268 $ 2,486 $ 838 3 Storage Software - EMC, IBM, HP, NetApp 31 $ 2,329 $ 5,433 $ 1,379 $ 2,009 4 Storage Software - Other 32 $ 902 $ 2,105 $ 523 $ 1,547 5 6 7 8 Server Storage (of which HP 38% and Dell 25%) 33 $ 10,125 $ 27,333 Storage purchased directly from ODM by Public Cloud and Private Cloud 34 $ 2,300 $ 8,086 Storage Software - Scale Out Object Based Software Defined Platform 35 $ 1,197 $ 4,566 Basic Cloud Storage (StaaS) - excludes SaaS like Box, Office 365 or Gmail 36 $ 1,500 $ 5,063 9 Symantec Software Storage 37 $ 2,097 $ 1,114 10 Tape Drive and Automation $ 1,800 $ 1,800 11 Storage Professional Services 38 $ 35,500 $ 31,101 TOTAL $ 93,071 $ 98,961 29 Latency-sensitive: inspired from EMC High-End last 3 quarters result (roughly -10% year-over-year). We assume -5% per year in 2015 and 2016, and -10% per year thereafter. Capacity-driven: assumes -5% per year in 2015 and 2016, and -20% per year thereafter. 30 Latency-sensitive: includes All Flash Array and Converged storage. We assume +5% per year as enterprise migrate to these efficient forms of storage. Capacity-driven: same as above. 31 We assumed same trend as Storage Systems of leading vendors (line 1). 32 Latency-sensitive: We assumed -6% per year for 2015 and 2016, and -10% per year thereafter as we see little use of storage software in conjunction with newer forms of latency-sensitive storage. Capacity-driven: We have assumed -5% per year, since a lot of server side backup software will continue to be relevant. 33 See Section II of this document 34 See Section II of this document 35 See Section II of this document 36 Inspired from Gartner. Assumes +20% per year in 2015, 2016 and 2017, accelerating to +25% per year afterwards 37 Assumes -10% per year since much of Symantec software portfolio becomes irrelevant with Software Defined Storage 38 Assumes +4% per year in 2015 and 2016, then stable in 2017 and 2018, then decreasing 10% per year in 2019 and 2020. 11 Scality 2014
Transition from Traditional Storage to Software Defined $70,000 $60,000 $50,000 $40,000 $30,000 $20,000 $10,000 $0 2014 2015 2016 2017 2018 2019 2020 Basic Cloud Storage Software Defined Storage Traditional Storage Systems (appliance) and Storage Software Overall, we expect the storage industry to continue to grow about 3 percent per year in 2014 and 2015, but then to stay stable at just below $100 billion. The traditional storage systems, as described above (line 1 to 4), decrease from $38 billion to $20 billion. In the same time, software defined storage for capacity-driven storage will grow from $13 billion to $40 billion, of which $27 billion are server revenues opportunity for OEM, and $4.5 billion is revenue opportunity for software vendors like Scality. IV. Bottom-up Approach from Vendor Revenues Another approach to market sizing is to look at what solutions SDS software like Scality would replace and how much revenue they make. For this approach Scality purchased from IDC the detail of revenue per product per quarter for the period Q3 2012 to Q2 2014, which makes up the $37 billion of Storage System revenue corresponding to lines 1, 2 and 5 in the previous section. This does not include the Storage Software revenue, so by definition, Scality and its software competitors are not included in the study. This study does not even include spend from entities who would buy the hardware directly from an ODM like Seagate, and then deploy Scality or an open-source software to transform it into a bona fide storage platform. Thus, a Scality deal like Los Alamos National Laboratory would not be included in the study. Still, the study is relevant since many of the systems Scality would replace or compete with, like Atmos, Isilon, DDN WoS, NetApp FAS are included in this study. We looked at what the $37 billion of storage system revenues are composed of and how much of a role Scality plays. Storage systems revenue is composed of external storage and internal storage, defined as follows by IDC: Internal storage includes the value of storage enclosed within application servers that contain three or more mass storage devices (HDD or SSD). It does not include the entire value of the server; rather, it includes the value of the mass storage devices, storage controllers, and other associated storage components. External storage encompasses all enterprise storage systems outside of the server enclosure. 12 Scality 2014
Scality s RING is deployed on internal storage systems. Typically, such systems need software to be used in the data center. When customers chose Scality, we typically charge between 25 percent and 30 percent of the server value. Traditional server values are 120 percent of the number reported in IDC study, because the IDC study does not include the entire value of the server. IDC reports a total of $8 billion for a 12-month period. This translates into a software opportunity for Scality of $8 billion x 1.2 x 0.25 = $2.4 billion. Scality often competes with external storage systems, but not all of them are relevant: We disregard storage systems which have a sale price below $50,000 (we focus on petabyte scale deployments) We disregard all flash systems The relevant revenue per vendor is listed below: Sum of Value (USD in millions) Quarter Vendor 2013 Q3 2013 Q4 2014 Q1 2014 Q2 DataDirect Networks 1 4 2 2 Dell 110 105 107 118 EMC 918 1,146 793 827 Hitachi 438 493 439 335 HP 315 395 290 373 Huawei 18 34 8 24 IBM 439 620 292 496 NetApp 539 573 644 569 Oracle 9 15 Grand Total 2,778 3,371 2,585 2,759 That s a total of $11.5 billion. When customers switch from a traditional system to SDS: Typically, they expect 25 percent reduction in cost Today, Scality is typically worth 30 percent of the total solution, but we expect this number to go down to 25 percent Which means that the Scality market opportunity opportunity is 18.75 percent of the traditional system revenue (100*0.75*0.25). The transformation of whole storage systems, as described above, into software defined storage creates a $2.1 billion opportunity. The internal storage opportunity of $2.4 billion, added to the external storage opportunity of $2.1 billion, equals $4.5 billion, to which opportunities that are not covered by the scope of the study must be added such as software opportunities arising from ODM sales to large public clouds. 13 Scality 2014
V. Bottom-up Approach from the Number of Potential Customers For forecasting purposes, we assumed that: Entities who spend over $1 million in storage per year are likely to move to SDS by 2020 Customers who deploy SDS at scale will spend an average of 60 percent of their storage for SDS, 25 percent of which for the software. According to IDC, in the US alone, and excluding government, research lab and education, there are 2,416 entities that spend between $1 million and $10 million in storage annually, 169 entities spending between $10 million and $100 million, and 18 entities spending over $100 million annually. This aligns with Gartner s Worldwide Datacenter Forecast, 39 which shows in the US alone: 2,300 datacenters with 26 to 100 racks, another 2,309 datacenters with 101 to 500 racks, and 598 datacenters of over 500 racks. Worldwide, Gartner forecasts that by 2018 there will be 22,070 datacenters in the world that are larger than 26 racks, and they will spend in aggregate $30 billion in storage systems per year. This leads to a business opportunity of $30 billion x 0.6 x 0.25 = $4.5 billion. This also implies that the average revenue per site for Scality is $200,000 per year, which is consistent with our observations. 39 Gartner: Forecast: Data Centers, Worldwide, 2010 2018, published August 2014 G269529 14 Scality 2014