Original title: “The Force Research | Filecoin storage revolution and ecological value, where does the sword point? (under)”
Author: Tony
Artificial intelligence (AI) has now encountered a big bottleneck, and that is data. The prerequisite for the development of artificial intelligence is to have a huge amount of data, but the current data owners do not have enough motivation to provide these data, especially the extremely valuable private data.
-Zhang Shousheng, tenured professor and physicist at Stanford University
Zhang Shousheng
Looking back at Filecoin’s commercial landing exploration series, we compare Filecoin with traditional projects Curve, HDFS, Ceph, and GlusterFS horizontally, and make vertical comparisons with blockchain projects Arweave, Crust, Burst, Bluzelle, Storj, and Sia, and explore in multiple dimensions. The commercial development path of HDFS and Ceph.
This time we continue to explore the commercial landing of Filecoin with Zhang Shousheng’s questions. This article will compare Filecoin and GlusterFS, and then analyze the market value of their respective landing directions.
3) GlusterFS: a fan of the native protocol
Gluster
On October 4, 2011, Red Hat spent $136 million to acquire Gluster. On April 30, 2014, Red Hat acquired Ceph enterprise product provider Inktank for $175 million. On October 28, 2018, IBM purchased all the outstanding shares of Red Hat for US$190 per share, with a total equity of approximately US$34 billion.
GlusterFS is an open source distributed file system with features such as high scalability, high availability, high performance, and horizontal scalability. Horizontal expansion can support several petabytes of storage capacity and handle thousands of clients at the same time. GlusterFS is also not perfect, and there are still some deficiencies in data storage security.
The main attack of GlusterFS is the native protocol of storage. Many cloud development service platforms support its applications, including Google Kubernetes and Red Hat OpenShift. GlusterFS is mainly applicable to the following three categories:
- Media: documents, pictures, audio, video
- Shared storage: cloud storage, virtualized storage, HPC (high-performance computing)
- Big data: log files, RFID (radio frequency identification) data
Based on the above analysis of the development path of Ceph, HDFS and GlusterFS, the final landing directions of distributed storage are: cloud computing facilities, data analysis software and storage native protocols. In addition, what other ground explorations are there for distributed storage? Let’s review the big event of distributed storage first.
Fourth, the trend of distributed storage
Source: “2020: What is the trend of storage development in the next decade?” 》, Production: IPFS Force Zone, 2020-10-27
Distributed storage evolution theory. The development process of distributed storage from 2010 to 2019 can be divided into three categories: distributed block storage, distributed file storage, and distributed object storage. As the relative market share of distributed storage continues to grow, the two will coexist. Among them, Compellent acquired by Dell and 3Par acquired by Hewlett-Packard have realized block-level virtualization, decoupling hard disks and RAID; VMware Virtual SAN (now named vSAN) and Nutanix are different from previous dedicated storage, which combines file system software with Server decoupling reflects different stages in the journey of distributed storage. Following that, distributed storage has spawned many new landing scenarios:
1) Decentralized computing market size in 2024: USD 16.23 billion. As a new computing model, decentralized computing provides users with powerful computing capabilities by integrating distributed resources such as storage, network, data, and hardware, and fundamentally realizes parallel processing, distributed computing and grid computing. To a certain extent, it can make up for the high cost, slow calculation and data leakage of centralized computing, and it also promotes the utilization of idle computing resources.
Source: Technavio, 2020-03
According to Technavio’s latest market research report, from 2020 (7.09 billion U.S. dollars) to 2024 (16.23 billion U.S. dollars), the global decentralized computing market is expected to grow by 9.14 billion U.S. dollars, expanding the market size at a compound annual growth rate of 32%. The decentralized computing market is brewing.
2) Decentralized hyper-converged market size in Q1 2020: US$3.9 billion. The decentralized hyper-converged framework is a further upgrade based on distributed storage. Its architecture realizes the integration of computing, storage, and network through software, and uses virtualization and decentralization to define the technical architecture of the data center. Its main solution is the agile response of new business architecture, big data processing and access.
Hyper-converged infrastructure, production: IPFS Force Zone, 2020-10-27
According to data from the International Data Corporation (IDC) Global Quarterly Converged System Tracking Report, the revenue of the global converged system market will increase by 4.5% year-on-year to US$3.9 billion in Q1 2020. It divides hyper-converged systems into three categories: certified reference systems and integrated infrastructure, integrated platforms, and hyper-converged systems. Among them, hyper-converged systems accounted for 50.9% of the entire converged system market in the Q1 quarter of 2020, nearly US$2 billion.
Global quarterly market share of converged systems, source: IDC, 2020-06-18
3) Intelligent + storage market scale in 2024: US$3.45 billion. There are two main pieces of intelligence + storage: storage intelligence and storage empowering intelligent AI.
Storage intelligence
What I hope to achieve is to predict future events based on business load, operation and maintenance management, etc., and to dynamically adjust the storage resource pool based on the predicted results. The storage resource allocation method is very suitable for the concept of distributed storage mining pools, because mining pools are inherently suitable for flexible allocation adjustment methods. As long as they are matched with mature AI technology, the resource allocation function that increases and decreases can be realized.
Storage empowers intelligent AI
It is based on the basic functions of AI preparation, training, inference, and archiving, and storage to meet the needs of different AI scenarios, including performance and response.
In short, the integration of the two solves the storage, protection and optimization of data, and realizes low-cost and high-efficiency storage.
Global artificial intelligence-storage market share budget, source: marketsandmarkets.com, 2019-12
The global artificial intelligence-storage market is expected to grow from US$10.4 billion in 2019 to US$34.5 billion in 2024 (compound annual growth rate of 27.1%). Among them, the Asia-Pacific region maintained the highest compound annual growth rate with the artificial intelligence-storage market.
4) Docking between private clouds, mutual penetration of public and private clouds
The essence of private cloud docking is the information exchange of the B-end enterprise, which is mainly to realize storage mining pooling, automation, API bridging, and the private cloud management platform to drive the creation, adjustment, optimization and even recovery of storage resources on demand.
The emerging public and private cloud interactive market requires blockchain to realize desensitization and monetization. The essence of the mutual penetration of public and private clouds is the communication between the B-side enterprise and the C-side. Because the early public cloud was mainly for the C-side (or small B), with the deepening and popularization of cloud computing, the B-side and the C-side need information interaction. However, for the consideration of various factors such as privacy protection, security, government compliance, management, compatibility, inertia, migration cost, etc., the storage market urgently needs the birth of distributed storage to ensure the interaction of desensitized information and the privacy of information protection.
Zhang Shousheng found the answer to the question before the article in the article “Blockchain technology is a new turning point in the Internet world”: “Once there is a blockchain, if the created data can be valued and agreed, it will form a The big data market enables artificial intelligence to go further.”
Blockchain can not only value and formulate private data, but also make the development of AI go further, such as the storage of AI operation records and the desensitization of data. In general, blockchain + storage can realize the use of valuable data, and AI makes the whole process more intelligent.
5) The original protocol empowers the unlimited potential of entities. The distributed storage native protocol mainly provides the most fundamental storage architecture for various applications, facilitating the connection between the protocol layer and the application layer. From the perspective of building a house, the native protocol is mainly a tool for building a house frame, and the application layer is mainly for decorating the house to meet people with different preferences. The scale of this market may be larger, but everything related to storage may be a real demand, such as medical data storage, financial data storage, and large file cold storage.
Source: marketsandmarkets.com, 2020-09
Among them, the global cloud storage market is expected to grow from USD 50.1 billion in 2020 to USD 137.3 billion in 2025 (compound annual growth rate of 22.3%). The growth of the cloud storage market continues to grow as the amount of enterprise data soars, providing remote users with ubiquitous data and needs. Among them, it covers vertical areas such as telecommunications, healthcare and life sciences, media and entertainment, consumer goods and retail , Manufacturing, government and public sectors, energy and utilities, and others (education and travel, etc.).
5. How will Filecoin interact?
According to Coinmarketcap data, Filecoin’s current total market value is about 800 million US dollars. The above several fields can be used as the exploration direction of distributed storage. Although these are in the embryonic stage and the annual market value is only several billion US dollars, the future potential is unlimited. In general, there are three directions in which Filecoin chimeric entities can be implemented:
Filecoin storage price, source: storage.codefi.network, 2020-10-20
Storage-oriented, technology supports more directions. In the early days, Filecoin accumulated storage capacity through cost-effective services (Filecoin cost 0.052 US dollars/G/Y, Amazon S3 0.276 US dollars/G/Y). Only after the data is accumulated can the value of the data be reflected. At the same time, the Protocol Lab is cooperating to develop four full nodes (go-filecoin, lotus, fuhon, and forest maintained by the IPFS Force Zone) to improve Filecoin network security and meet the greater needs of the Filecoin ecosystem.
Free market development, parallel exploration at multiple points. After the structure is stable, the global team can build applications based on the Filecoin network characteristics, and innovators can develop different products based on their understanding of the market, and conduct multi-domain and multi-dimensional exploration.
Silent data urgently needs blockchain to awaken value. Zhang Shousheng believes that private data (that is, silent data) may have the greatest value, so he is committed to using blockchain + AI to wake up silent data. At present, most of the silent data in the market is not used, and its value is immeasurable. Once the data is used, it may be a new change to the world.
