SDH was designed as to be an answer to solve the flaws and problems of its predecessor plesiochronous digital hierarchy (PDH), but SDH itself will be replaced newer technologies and other demands that come about with the evolving internetwork. Therefore, aspects of SDH that could be categorised as strengths may also be a weakness when viewed from emerging technologies.

Summarizing the techonolgy
- Some bytes, called the payload, of pre-defined length are grouped into a container, with each of them getting some overhead information mapping into a virtual container (VC).
- Depending on the size of the VC, the smaller ones are grouped together, aligned, into tributary units (TUs), who are then combined (=multiplexed) to tributary unit groups (TUGs).
- They then are multiplexed into a VC-4, the largest possible virtual container. The largest defined container, being 139264 kbits/s, goes straight into a VC-4 without the aligning and multiplexing.
- The contents of this VC-4 is poured into an Administrative Unit, and mapped into an Administrative Unit Group (AUG).
- Well, finally these AUGs go into an STM frame having in total 9x270 bytes, racing at 155.52 Mbit/s through the fibre (or STS if you're thinking about SONET. STM stands for Synchronous Transfer Module, in case you're interested.).
All this mixing, matching and combining together can make it up to an aggregate bandwidth of 9953.280 Mbit/s with an STM-64 (= 64 STM-1 frames combined). To make a bit of sense out of such a number: that could carry 120000 telephone calls (or faxes), or 4000 low-scan video pictures, or 64 broadcast TV channels... nice!

General comments
SDH solved problems encountered with PDH in having defined a worldwide standardisation as a first (of which SONET is technically speaking a subset, although SDH evolved from SONET), enabling better integration to achieve a global network demanded by globalisation of the economy and allowing for higher aggregate bandwidth than PDH.
Technically, it made the multiplex mountain of PDH obsolete with the introduction of add/drop multiplexers (ADMs) so that a path can be extracted from a frame and routed to another destination without the need of de- and re-multiplexing the whole frame. Third, it addressed the need for better services (like error performance and maintenance) by defined section and path overhead bytes in the frame and an embedded data communications channel. This extra information and reduced amount of equipment allows for greater flexibility and more sophisticated and centralized network management.

However, SDH is not heaven on earth. To keep the network synchronised and to minimise jitter, expensive high standard clocks in a distributed timing system are required.
Because equipment costs are high, research into eliminating SDH from the network is carried out; for example IP (over ATM) on WDM, which results in a simpler transmission technique with a lower overhead.
SDH still requires allocation of resources like connecting ports in a digital cross-connect (DXC), but it would be more ideal when the management system only would have to bother with setting policies and leave the implementation intelligence to the network elements.

Summarizing, the advantages of SDH are network simplification, its survivability, software control, bandwidth on demand and standardisation. Disadvantages are keeping the network synchronised, high cost and management functionality not meeting the new demands.

Brattli, T. SDH tutorial. to topic11. 1999.
Hawker, I., Hill, G. and Taylor, I. UK core transmission network for the new millennium. BT Technology Journal, Vol 18 No 3. July 2000.
Moss, N.. Synchronous Digital Hierarchy. Systems and Processes, T305 syllabus Block 2. 1999. pp 6-71.