— 领先的中文存储网络媒体 | 51CTO旗下网站

方案案例 > 大数据 > 正文
作者: 佚名 2014-06-06 13:07 【】

Challenge: Limitations of Legacy File-Based Storage

The established paradigm of hardware-centric, file-based systems is becoming increasingly problematic as the fundamental nature of enterprise storage shifts. Rapid growth of large-scale, data-driven applications is highlighting inefficiencies in traditional data centers, whose legacy architectures are not well-adapted to the highly distributed and capacity-optimized workloads presented by this burgeoning wave of unstructured data and applications.

Unfortunately, current data centers still rely on a convoluted storage architecture that uses multiple layers of software and hardware stacked together to provide a data path between two poorly-compatible systems: an object-oriented application layer and a hardware layer (spanning HDDs, SSDs and tape) based on block storage. This tortuous path from application to storage entails numerous layers of manipulation from databases down through POSIX interfaces, file systems, volume managers and drivers; data then passes over Ethernet, through Fibre Channel, into RAID controllers, SAS expanders and SATA host bus adapters.

Moving information through these multiple layers is not only inefficient, it is also unnecessary. The majority of today’s mass scale object applications do not need either file semantics or a file system to optimize space management on a device. Modern applications only need object semantics (for example, write the whole thing, read the whole thing, delete the whole thing, refer to it by a handle chosen by the client and cluster manager), not where data resides on a given device.

Solution: Object Storage With Ethernet-Direct HDD Technology

Ethernet is the undisputed fabric of data centers and application traffic, and has unsurprisingly emerged as the backbone of storage infrastructure (as evidenced by rapid growth of virtual server and desktop infrastructure, widespread adoption of Amazon Web Services, etc.). This makes Ethernet the ideal foundation for a new object-based platform (the Seagate Kinetic Open Storage platform) that vastly streamlines and simplifies the way applications communicate with storage devices.

Newisys EDA-4605 Ultra-Dense Ethernet Drive Storage Array

The Newisys EDA-4605 is paving the way for groundbreaking developments in Ethernet-direct hard disk drive (HDD) technology. In fact, the first HDD product to leverage the EDA-4605’s remarkable capabilities is the innovative new Seagate Kinetic Open Storage HDD. Optimized for object storage, the EDA-4605 provides an ultra-efficient storage platform for cloud and big data deployments, and is ideal for scale-out and distributed storage solutions.

Unlike traditional storage boxes, the EDA-4605 employs redundant Ethernet fabrics that connect directly to the HDDs, eliminating multiple layers of overhead and enabling new levels of storage scalability. Whether deployed in cloud installations, big data environments or traditional data centers, the EDA-4605 delivers object storage with unprecedented efficiency.

With up to 60 3.5-inch Seagate Kinetic Open Storage HDDs per 4U enclosure, the industry-unique Newisys EDA-4605 is an ultra-dense, space- and power-saving storage solution that significantly reduces deployment and management complexity. The Newisys EDA-4605 fits nicely into a standard 19-inch wide, 1-meter deep rack and provides a flexible 15 HDDs/U object storage building block. With 4TB HDDs, this can deliver 2.4 petabytes per standard 42U rack, and can easily scale out beyond that.

 Figure 2. Rear of Newisys EDA-4605 Ultra-Dense Ethernet Drive Storage Array

Key Benefits

Optimized for Ethernet-direct HDDs, such as the innovative Seagate Kinetic Open Storage platform

Ideal building block for object storage deployments

Reduces complexity and improves efficiency for cloud and big data object storage installations

Full-featured, highly available, high-performance object storage

Product Features

Four 10GbE connections to data center

Redundant, 1GbE connections to each of 60 HDDs

Dual-redundant, hot-pluggable Ethernet switch module (ESM) management

Dual-redundant, hot-pluggable, high-efficiency power supply units (PSU) and fans

Redundant, hot-pluggable system blowers implemented in the PSUs

Modular design to increase product configuration flexibility

Standard chassis customization and branding available

Greater Scale-Out, Lower TCO Are Key Benefits of Newisys EDA-4605

The Newisys EDA-4605, based on the Seagate Kinetic Open Storage platform is uniquely optimized for explosive-growth, scale-out data centers. By disaggregating storage from compute, the platform’s architecture allows data center operators to simply add more storage as capacity needs grow. This technological advantage facilitates denser storage racks, which can decrease TCO in a number of different areas:

Lower capital expenditures—Storage servers can be removed from the data center, resulting in lower capital expenditures when building out data centers. Alternatively, for a given level of capital expenditure, customers can allocate much greater storage capacity in the same physical space.

Labor—Removing the need for storage servers reduces the number of technicians required to maintain them. In addition, the higher-density storage enabled by the platform can potentially reduce the number of technicians a data center needs to employ in general, leading to significant labor savings.

Power consumption—Eliminating the storage server tier, combined with more efficient rack density, enables fewer racks to support the same quantity of storage and thus reduces energy consumption.

Uptime/technician error—The platform’s greater reliability features (for example, automatic replication and failover) help to reduce the number of storage management-related errors in the data center. In addition, in-drive error management promises to reduce the number of major incidents that require a technician’s attention.

The increase in rack density enabled by the Seagate Kinetic Open Storage platform provides another compelling cost benefit—a significantly smaller physical footprint for the data center, which directly translates into lower costs. This is particularly critical in cases where availability of physical real estate is constrained (for example, data centers located in co-location facilities).

The platform’s precise impact on storage TCO will vary according to a number of factors (number of HDDs behind each existing storage server, real estate cost criteria, etc.) specific to a given data center, but preliminary indications show potential for significant capex and opex savings.

标签:大数据 混合云 数据保护 云存储 服务器虚拟化