Data centers are the epitome of infrastructure automation, and their modern manifestation—cloud—provides an almost magical platform for its users. To construct clouds, separation of concerns into layers of abstraction, like network overlays and service API encapsulations, help enable service agility and innovation. But do these layers curb complexity, or merely mask it?
In the enterprise, there isn’t a hotter battlefield for new technology than the data center. It sits at the center of cloud, so it is a natural point of emphasis in virtually every enterprise networking strategy. It garners a big portion of total enterprise networking spend, so it is an area of importance to virtually every networking vendor. And it is ultimately the place-in-network (PIN) on top of which most applications run, so it is a target for many technological innovations.
So how can it be that so many people get the data center discussion wrong?
It is well understood that the technology that underpins the data center is changing. In fact, it’s not just changing—it’s accelerating. And that acceleration and change ultimately drives changes in how leaders think about how they design, build, and operate their infrastructure.
In a previous blog, I introduced some research that PwC conducted on the changing buying patterns around data center networking. The research was conducted over several months, featuring 35 live interviews and 235 unique survey responses coming primarily from CIOs and VPs of IT from a mix of global companies.
The role of IT is changing. As IT used to be viewed as primarily a services organization, modern enterprises of even moderate size and sophistication are relying on IT as a key strategic contributor. With the shift in responsibility, there is a commensurate shift in IT leadership priorities. This shift is the focus of recent research led by PwC.
Today enterprise data centers face huge pressure to provide increased bandwidth for Digital Transformation, Big Data and IoT applications. The increasing adoption of virtualized workloads (VMs and containers) and ongoing transition from on prem to off prem cloud services has resulted in significant strain on the spine-leaf architectures. The higher bandwidth at the access layer (10/40 GbE) has driven the need for greater upstream bandwidth (100GbE).
To account for this need, Juniper is launching QFX5110 switch – the latest addition to the QFX5100 family of switches that provides 100GbE uplink access to the aggregation layer along with features designed to optimize today’s agile data centers. Based on Broadcom’s Trident 2+ chipset, the QFX5110 will be available in two form factors:
QFX5110-48S: Compact, 1U 10GbE/100GbE data center access switch with 48 small form-factor pluggable plus (SFP+) transceiver ports and 4 QSFP28 ports, which can be configured as 4x40GbE or 4x100GbE ports.
QFX5110-32Q: Compact, 1U 40GbE/100GbE data center access and aggregation switch offering up to 32 QSFP+ ports, or 20 QSFP+ ports and 4 QSFP28 ports.